article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
this work is dedicated to prof .
jerzy zabczyk , our master and teacher .
let @xmath1 be a real separable hilbert space .
in this paper we are concerned with the ergodic behaviour of the semilinear stochastic equation @xmath2 driven by , possibly cylindrical , lvy noise @xmath3 .
we assume that @xmath4 is a generator of a @xmath5-semigroup @xmath6 of bounded linear operators on @xmath1 and @xmath7 is a lipschitz mapping . under the conditions specified in section [ main_result ]
equation has a unique solution @xmath8 for every @xmath9 .
our main question is to provide conditions for the exponential convergence of transition measures @xmath10 to a unique invariant measure @xmath11 in the norm of total variation .
this question has been thoroughly investigated in the case when @xmath12 with a certain cylindrical wiener process @xmath13 and @xmath14 a linear operator , see for example @xcite . in those works
more general , non - lipschitz drifts @xmath15 were permitted .
the two main ingredients for the proofs were the strong feller property and topological irreducibility of the transition measures @xmath16 .
+ in this paper we extend the aforementioned approach to equation with general ( subject to some natural conditions ) lvy noise @xmath3 and bounded lipschitz drift @xmath15 .
exponential ergodicity for equation was investigated recently in @xcite . in theorem [ th1 ] of section [ main_result ]
we show that if the semigroup @xmath17 is exponentially stable and compact and the measures @xmath16 are strong feller and irreducible then the convergence to the invariant measure holds in the total variation norm and moreover the rate of convergence is exponential . the proof is based on an idea developed in @xcite . in section [ examples ]
we discuss three examples .
in subsection [ stable ] we apply our general result to equation ( [ 1 ] ) with the noise @xmath3 being a symmetric @xmath0-stable cylindrical noise . in corollary [ cor1 ]
we show that theorem [ th1 ] yields a strengthening of the similar result obtained in @xcite and of an earlier result in @xcite . in the proof we use the fundamental results of priola and zabczyk proved in @xcite . in subsection [ sec - gaussian ]
we consider lvy noise with both , the jump part and the gaussian part , non - vanishing .
finally , in section [ singular ] we consider the stochastic heat equation driven by a mixture of a cylindrical symmetric @xmath0-stable noise and a singular wiener noise with unbounded covariance operator .
* notations . * in what follows ,
the norm in the hilbert space @xmath1 is denoted by @xmath18 .
the norm in any other banach space @xmath19 will be denoted as @xmath20 .
the hilbert - schmidt norm of an operator @xmath21 will be denoted as @xmath22 .
we start with the set of conditions that are assumed to hold throughout the paper .
we assume that @xmath23 is a lvy process defined on a stochastic basis @xmath24 satisfying the usual conditions and with values in a hilbert space @xmath25 , such that @xmath26 with continuous and dense embedding . more precisely , we assume that @xmath27 where @xmath28 is a fixed vector , @xmath13 is a standard cylindrical wiener process in @xmath1 , the process @xmath29 is a @xmath25-valued , pure jump lvy process , independent of @xmath13 , and @xmath30 is a selfadjoint and nonnegative operator .
+ we will assume that the semigroup @xmath17 enjoys the following properties .
[ h1 ] 1 . for every @xmath31 $ ]
, the operator @xmath32 has an extension ( still denoted by @xmath32 ) to a bounded operator @xmath33 .
we have @xmath34 for each @xmath35 the operator @xmath36 extends to a bounded operator on @xmath1 and @xmath37 [ h2 ] for every @xmath31 $ ] the stochastic integral @xmath38 is a well defined @xmath1-valued random variable defined as a limit in probability of the corresponding stieltjes sums ( see e.g. @xcite and @xcite ) .
moreover , there exists @xmath39 such that @xmath40}\e\left|\int_0^ts_sdy_s\right|^p<\infty,\ ] ] and @xmath41 the remark below provides a convenient sufficient condition for hypothesis [ h2 ] to hold .
it is proved in section [ proofs ] .
[ 22a ] if @xmath42 and @xmath43 for a certain @xmath39 then hypothesis [ h2 ] is satisfied with this @xmath44 .
finally , we impose the standard lipschitz condition on the nonlinear term @xmath15 .
[ h3 ] the function @xmath7 is lipschitz continuous and bounded .
[ pr1 ] assume hypothesis [ h2 ] and let @xmath45 then the following hold
. 1 . for any @xmath46 there exists a constant @xmath47 such that @xmath48 2 .
the process @xmath49 is continuous in @xmath44-th mean .
the process @xmath49 admits a predictable modification .
[ rem1 ] clearly , proposition [ pr1 ] holds true for a more more general process @xmath50 with @xmath51}\e\left|\int_0^ts_sdz_s\right|^p<\infty,\ ] ] by , and .
we will always consider a predictable modification of @xmath52 .
let us recall that the process @xmath53 is a mild solution to equation if it is @xmath1-valued , predictable and @xmath54 we will use the notation @xmath55 [ pr2 ] assume hypotheses [ h1]-[h3 ] .
then for every @xmath9 equation has a unique mild solution @xmath53 .
the process @xmath53 is markov and feller in @xmath1 , @xmath56 and the process @xmath53 is @xmath57-continuous in @xmath44-th mean .
now we can formulate the main result of this paper . in the theorem below
we denote by @xmath58 the total variation norm of a signed borel measure @xmath59 on @xmath1 . for the definitions of the strong feller property and open set irreducibility the reader may consult @xcite .
[ th1 ] assume hypotheses [ h1]-[h3 ] .
moreover assume that the semigroup @xmath17 is compact and exponentially stable .
assume also that the solution @xmath60 to is strongly feller and open set irreducible .
then there exists a unique invariant measure @xmath11 for and there exist constants @xmath61 and @xmath62 such that @xmath63 [ rem2 ] let @xmath64 be the solution to starting from an @xmath65-measurable random variable @xmath66 such that @xmath67 .
then inequality extends by standard arguments to a more general one @xmath68 where @xmath69 is the law of the random variable @xmath64 .
in this section we apply theorem [ th1 ] to equation driven by a cylindrical symmetric @xmath0-stable noise @xmath3 with @xmath70 , introduced by priola and zabczyk in @xcite and later studied in @xcite and @xcite .
let @xmath71 and let @xmath72 be a cons in @xmath1 .
we will identify @xmath1 with the space @xmath73 of square - summable sequences .
let @xmath74 be an arbitrary sequence of positive real numbers .
then we can define the hilbert space @xmath75 of all sequences @xmath76 such that @xmath77 and we will denote by @xmath78 the corresponding inner product . + the equation @xmath79 is studied in @xcite and in @xcite in the space @xmath80 under the set of assumptions listed below .
[ hpz ] 1 .
let @xmath4 be a selfadjoint operator @xmath81 such that @xmath82 with @xmath83 2 . for @xmath70
we define a cylindrical @xmath0-stable process @xmath84 where @xmath85 is a sequence of real - valued symmetric and independent @xmath0-stable processes and @xmath86 is a sequence of positive reals such that @xmath87 3 .
the mapping @xmath7 is lipschitz continuous and bounded .
4 . there exist constants @xmath88 and @xmath89 such that @xmath90 let for a fixed @xmath91 , @xmath92 clearly , @xmath1 is continuously and densely imbedded into @xmath93 .
+ we shall show that hypothesis [ hpz ] implies the assumptions of proposition [ pr2 ] and of theorem [ th1 ] .
+ first , note that @xmath3 is an @xmath93-valued lvy process .
indeed , by proposition 3.3 in @xcite , @xmath3 is a lvy process with values in @xmath94 with @xmath95 such that @xmath96 hence with values in @xmath93 by .
+ next , the operator @xmath4 with the domain @xmath97 , where @xmath98 generates on @xmath1 a symmetric compact semigroup @xmath17 with @xmath99 in particular , @xmath100 therefore , @xmath101 fix @xmath35 and let @xmath102 .
then @xmath103 since @xmath104 , @xmath105 and yields @xmath106 and the proof of part ( 1 ) of hypothesis [ h1 ] is complete .
+ since @xmath107 and @xmath108 all the conditions of hypothesis [ h1 ] are satisfied .
+ next , if parts ( 1 ) and ( 2 ) of hypothesis [ hpz ] hold then it follows from proposition 4.4 of @xcite that for any @xmath35 the stochastic integral @xmath109 takes values in @xmath1 , hence the first part of hypothesis [ h2 ] is satisfied and additionally for any @xmath110 @xmath111 where the constant @xmath112 depends on @xmath44 only .
we note that the last estimate is an immediate consequence of theorem 4.6 in @xcite .
note also that by the rhs of is finite uniformly in @xmath35 , hence our condition is satisfied with @xmath113 moreover , using the dominated convergence we find that the rhs of tends to 0 as @xmath114 which yields our condition .
finally , all the conditions of hypothesis [ h2 ] hold .
+ hypothesis [ h3 ] is identical with part 3 of hypothesis [ hpz ] .
+ thus we have proved that under assumptions ( 1)-(3 ) of hypothesis [ hpz ] , our hypotheses [ h1]-[h3 ] are satisfied and thereby as an immediate consequence of proposition [ pr2 ] and theorems 5.5 and 5.7 in @xcite we obtain [ cor1 ] assume that hypothesis [ hpz ] holds .
then for every @xmath9 there exists a unique mild solution @xmath53 to equation which is an @xmath1-valued markov process , irreducible and strong feller .
moreover , for any @xmath110 @xmath115 finally , taking into account and we note that by hypothesis [ hpz ] all the assumptions of our theorem [ th1 ] are satisfied and we obtain immediately the following strengthening of theorem 2.8 in @xcite and of theorems 2.2 and 2.3 in @xcite .
[ cor2 ] under hypothesis [ hpz ] there exists a unique invariant measure for and for any @xmath110 there exist constants @xmath61 and @xmath62 such that @xmath116 [ 34 ] ascertainment is proved in theorem 2.8 in @xcite under hypothesis [ hpz ] and an additional assumption that for some @xmath117 @xmath118 let us note , that ensures that the solution @xmath60 to takes values in a linear subspace compactly imbedded into @xmath1 , while of hypothesis [ hpz ] is equivalent to the fact that @xmath60 is @xmath1-valued only ( comp . remark 2.3 in @xcite ) .
corollary [ cor2 ] strengthens also theorems 2.2 and 2.3 from an earlier paper @xcite . in that paper additionally to hypothesis [ hpz ]
the nonlinearity @xmath15 must be sufficiently small and only weak convergence to the invariant measure @xmath11 with exponential rate is obtained . in this section
we consider equation with @xmath119 , i.e. @xmath120 under hypotheses [ h1 ] and [ h2 ] . by the operator @xmath121 defined as
@xmath122 is nonnegative and trace class for every @xmath35
. we will also assume the following [ h3b1 ] @xmath123 let us recall that hypothesis [ h3b1 ] holds if and only if the solution @xmath124 ( gaussian ornstein - uhlenbeck process ) , corresponding to equation with @xmath125 and @xmath107 , is strong feller ( see section 9.4.1 of @xcite ) .
therefore , if hypothesis [ h3b1 ] holds then the solution @xmath60 to is strong feller as well by lemma 20.1 in @xcite .
we note also that by hypothesis [ h3b1 ] and the strong continuity of the semigroup @xmath17 the set @xmath126}q_s^{1/2}(h)\ ] ] is dense in @xmath1 for every @xmath35 .
hence , @xmath127 has a support on the whole of @xmath1 for every @xmath35 .
therefore @xmath128 is irreducible for every @xmath35 and @xmath9 by lemma 20.1 in @xcite .
thus we obtain the following immediate consequence of theorem [ th1 ] .
[ cor1 g ] assume hypotheses [ h1 ] , [ h2 ] and [ h3b1 ] and moreover assume that the semigroup @xmath17 is exponentially stable .
then equation has a unique invariant measure @xmath11 that satisfies . in particular ,
invoking remark [ 22a ] we obtain [ 2 g ] let @xmath42 and @xmath43 for certain @xmath39 .
moreover , assume and hypothesis [ h3b1 ] and that the semigroup @xmath17 is exponentially stable .
then equation has a unique invariant measure @xmath11 that satisfies . in this section
we will apply theorem [ th1 ] to the stochastic heat equation @xmath129\\ & \\ x_t(0)=x_t(\pi)=0,&t>0 , \end{array}\right.\ ] ] with an unbounded operator @xmath130 , vector @xmath28 , and a cylindrical @xmath0-stable process @xmath29 .
it is well known that equation can be rewritten as a version of an abstract equation .
indeed , let @xmath131 and @xmath132 then @xmath4 has eigenfunctions @xmath133 and eigenvalues @xmath134 we assume that for a certain @xmath135 @xmath136 then @xmath137 and therefore , by proposition 9.30 in @xcite hypothesis [ h3b1 ] holds .
note that holds for @xmath4 and @xmath138 defined above iff @xmath139 indeed , is satisfied iff @xmath140 which is equivalent to , comp .
proposition 9.30 in @xcite .
in particular , for @xmath141 the operator @xmath142 in is unbounded but condition holds .
+ let @xmath143 where @xmath144 is a sequence of real - valued , independent , symmetric and @xmath0-stable processes with @xmath70 .
we will assume that the sequence of non - negative real numbers @xmath86 enjoys the following property .
+ the set @xmath145 is infinite and there exists @xmath146 such that @xmath147 observe , that if @xmath148 then condition holds and by results in section [ stable ] hypothesis [ h2 ] is satisfied for @xmath149 . in particular ,
if @xmath150 then the process @xmath29 defined by and is a cylindrical noise taking values in the space @xmath151 ( see section [ stable ] ) . note that if @xmath152 then the process @xmath29 is degenerate .
let @xmath153 that is @xmath154 then for every @xmath46 @xmath155 and follows .
finally , as a consequence of corollary [ cor1 g ] we obtain the following [ 38 ] under assumptions - equation has a unique stationary distribution @xmath11 and for any @xmath110 exponential estimate holds .
if @xmath156 , then the process @xmath157 is an @xmath158-martingale and and follow easily .
+ let @xmath159 .
it has been proved in @xcite , see also @xcite for a more general result , that @xmath160 iff @xmath161 where @xmath162 is the lvy measure of @xmath29 .
we can represent @xmath163 where @xmath164 is a lvy process corresponding to the lvy measure @xmath162 restricted to the ball @xmath165 and @xmath166 is a lvy process corresponding to @xmath167 , the lvy measure @xmath162 restricted to @xmath168 , and moreover @xmath164 and @xmath166 are independent . therefore is satisfied iff holds for @xmath166 .
by @xcite we have @xmath169 where @xmath170 is the poisson measure corresponding to @xmath166 , in particular @xmath170 takes nonnegative integer values and hence @xmath171 for @xmath172 .
therefore as in @xcite we obtain the estimate : @xmath173 where the last inequality follows from and and hold .
we will consider only the case of @xmath174 which is less standard .
the proof for @xmath175 is similar .
we recall that for @xmath174 the space @xmath176 is a linear complete metric space with the distance @xmath177 _ proof of ( 1)_. for @xmath178 , @xmath179 we have ( e.g. @xcite ) @xmath180 where @xmath181 stands for the equality of probability distributions .
hence , invoking the fact that for certain constants @xmath182 and @xmath183 @xmath184 we find that for a fixed @xmath35 @xmath185 - 1}\e\left|\int_k^{k+1}s_sdy_s\right|^p+\left\|s_{[t]}\right\|^p\e\left|\int_0^{t-[t]}s_sdy_s\right|^p\\ & \le \gamma_pc_a^p\sum_{k=0}^{[t]}e^{p\delta k}<\infty , \end{aligned}\ ] ] and follows with @xmath186}e^{p\delta k}.\ ] ] in estimates follow from the triangle inequality and .
+ _ proof of ( 2 ) . _ fix @xmath46 .
then for @xmath187 @xmath188 hence , by triangle inequality and @xmath189 and by of hypothesis [ h2 ] , @xmath190 concerning @xmath191 observe that for @xmath192-a.e .
@xmath193 @xmath194 by the strong continuity of the semigroup @xmath17 .
moreover , for @xmath195 $ ] @xmath196 therefore , by part ( 1 ) of the proposition and the dominated convergence we obtain @xmath197 and the right - continuity of @xmath49 in @xmath44-th mean follows .
similarly , to prove left - continuity , fix @xmath35 and let @xmath198 .
then , using we obtain @xmath199 and @xmath200 therefore , by and @xmath201 and part ( 2 ) of proposition [ pr1 ] follows .
+ _ proof of ( 3 ) .
_ the chebyshev inequality and part ( 2 ) of proposition [ pr1 ] yield the stochastic continuity of the process @xmath49 .
since @xmath49 is @xmath202-adapted , part ( 3 ) follows from proposition 3.6 of @xcite .
we follow the proof of theorem 5.4 in @xcite . recall that we can choose a predictable version of the process @xmath52 .
uniqueness of solutions to is an easy consequence of the gronwall lemma . to prove existence observe first that under hypothesis [ h3 ] , for fixed measurable function @xmath203\to h$ ] and @xmath9 ,
the deterministic equation @xmath204 has a unique continuous solution @xmath205\to h$ ] , hence by part ( 3 ) of proposition [ pr1 ] there exists a continuous @xmath1-valued and @xmath206-adapted process @xmath207 such that @xmath192-a.s .
@xmath208 consequently , the process @xmath209 ,
@xmath46 , is a predictable solution to . by proposition [ pr1 ] and remark [ rem1 ] we have @xmath210 .
the markov property is proved in theorem 5.4 of @xcite and the feller property follows from the gronwall lemma .
_ we show first that there exist positive constants @xmath211 such that @xmath212 to this end we note that by hypothesis [ h3 ] @xmath213 and since the semigroup @xmath17 is exponentially stable , there exist constants @xmath214 such that @xmath215 therefore , using remark [ rem1 ] and for @xmath216 we obtain @xmath217}e^{-\kappa pj}\\ & \le c_a^pe^{-\kappa pt}|x|^p+c_a^pc_f^p\kappa^{-p}+\tilde{\gamma_p}c_a^p\la 1-e^{-\kappa p}\ra^{-1 } , \end{aligned}\ ] ] and follows .
+ _ step 2 .
_ in order to prove the existence of an invariant measure for equation we will use the method of @xcite and @xcite to show that the family @xmath218 is tight .
indeed , for @xmath219 @xmath220 fix arbitrary @xmath221 . since for any @xmath219 @xmath222 there exists a compact set @xmath223 such that @xmath224 next , by @xmath225 hence by the chebyshev inequality there exists @xmath226 such that @xmath227 where @xmath228 stands for the closed ball @xmath229 . since @xmath230 is compact , the set @xmath231 is compact in @xmath1 and @xmath232 for all @xmath219 .
finally , since @xmath17 is a compact semigroup , by lemma 6.1.4 in @xcite the operator @xmath233;h)\to h$ ] defined as @xmath234 is compact .
then , invoking the boundedness of @xmath15 we find that there exists a compact set @xmath235 such that @xmath236 let @xmath237 clearly , @xmath238 is a compact subset of @xmath1 and @xmath239 therefore , the set @xmath240 is tight and an invariant measure for exists by the krylov - bogoliubov theorem ( e.g. theorem 16.2 of @xcite ) .
+ _ step 3 .
_ to show we will use the following theorem 12.1 proved in @xcite .
since the process @xmath8 is strongly feller and irreducible , it remains to show .
fix @xmath242 and @xmath243 .
then there exists a compact @xmath244 such that @xmath245 . by the strong feller property all the measures in the family @xmath246 are equivalent , hence @xmath247 to prove we use a nice observation of @xcite : since @xmath228 is compact in the weak topology @xmath248 of @xmath1 , will follow from and the fact that the function @xmath249 it remains to prove . to this end , take a sequence @xmath250 , such that @xmath251 . since @xmath32 is compact for any @xmath35 we have @xmath252 next , by the lipschitz continuity of @xmath15
we obtain @xmath253 hence for a certain constant @xmath254 the gronwall lemma yields @xmath255 and thereby , using dominated convergence we obtain @xmath256 now , by the dominated convergence @xmath257 for any @xmath35 and any @xmath258 .
putting in @xmath259 , and @xmath260 by the strong feller property , we find that @xmath261 which implies since @xmath262 is metrisable .
da prato g. and zabczyk j. : * stochastic equations in infinite dimensions * , cambridge univ . press , cambridge , 1992 .
da prato g. and zabczyk j. : * ergodicity for infinite - dimensional systems * , cambridge univ . press , 1996 goldys b. and maslowski b. : exponential ergodicity for stochastic reaction - diffusion equations , in * stochastic partial differential equations and applications vii * , lecture notes pure appl . math .
245 ( 2006 ) , 115 - 131 , chapman hall / crc press jurek z.j . and smalara j. : on integrability with respect to infinitely divisible measures , _ bull .
acad . pol .
sci . , math .
_ 29 ( 1981 ) , 179 - 185 kruglov v.m . : on infinitely divisible distributions in hilbert space [ in russian ] , _ matematiceskie zamietki _ 16 ( 1974 ) , 585 - 594 maslowski b. : on probability distributions of solutions of semilinear stochastic evolution equations , _ stochastics stochastics rep . _ 45 ( 1993 ) , 17 - 44 maslowski b. and seidler j. : on sequentially weakly feller solutions to spde - s , _ rend .
lincei _ 10 ( 1999 ) , 69 - 78 priola e. and zabczyk j. : harmonic functions for generalized mehler semigroups , in * stochastic partial differential equations and applications vii * , lecture notes pure appl .
245 ( 2006 ) 243 - 256 , chapman hall / crc press | we study convergence to the invariant measure for a class of semilinear stochastic evolution equations driven by lvy noise , including the case of cylindrical noise . for a certain class of equations
we prove the exponential rate of convergence in the norm of total variation .
our general result is applied to a number of specific equations driven by cylindrical symmetric @xmath0-stable noise and/or cylindrical wiener noise .
we also consider the case of a `` singular '' wiener process with unbounded covariance operator .
in particular , in the equation with diagonal pure @xmath0-stable cylindrical noise introduced by priola and zabczyk we generalize results in @xcite . in the proof we use an idea of maslowski and seidler from @xcite . |
galaxy formation and evolution remain important issues in our understanding of the early epoch of the universe .
fundamental steps towards these issues would be the reconstruction of the chemical histories of galaxies and the metal census of the universe .
they are investigated both via the traditional approach of observing the galaxy starlight , and via the absorption line systems detected along the quasar lines of sight independently of their distance , luminosity and morphology .
the absorption line systems , in particular , provide a very accurate observational method to measure the neutral gas and metal content of the universe up to very high redshifts , @xmath8 ( e.g. lu et al .
1996a ; prochaska , gawiser & wolfe 2001 ; dessauges - zavadsky et al . 2001a ; songaila & cowie 2002 ) .
the quasar absorption line systems are divided into three classes according to their neutral hydrogen column density : the ly@xmath2 forest with @xmath9 ( ) ranging from @xmath10 to @xmath11 cm@xmath1 , the lyman limit systems ( lls ) with @xmath9 ( ) @xmath12 cm@xmath1 , and the damped ly@xmath2 systems ( dlas ) with @xmath9 ( ) @xmath13 cm@xmath1 ( wolfe et al . 1986 ) .
they thus probe different media from voids to halos and disks of both dwarf and normal ( proto-)galaxies ( prochaska & wolfe 1998 ; ledoux et al .
1998 ; haehnelt , steinmetz & rauch 1998 ; matteucci , molaro & vladilo 1997 ; jimenez , bowen & matteucci 1999 ; calura , matteucci & vladilo 2003 ) .
the dlas , in particular , are a cosmologically important population , since they dominate the neutral hydrogen content of the universe ( e.g. wolfe et al .
1995 ; storrie - lombardi & wolfe 2000 ) and are believed to be the progenitors of the present - day galaxies .
[ cols="<,^,<,<,<,^,^ , < " , ] @xmath14 computed for the sample of dlas only . +
* references : * this paper ; dessauges - zavadsky et al .
( 2001a ) ; lopez et al .
( 1999,2002 ) ; lu et al .
( 1995,1996b ) ; prochaska & wolfe ( 1996,1997,1999 ) ; prochaska et al .
( 2001 ) ; proux et al .
( 2002b ) it should be emphasized that the observed low - ion transition line profiles extend over a very large velocity interval of @xmath15 km s@xmath16 , from @xmath17 to @xmath18 km s@xmath16 in velocity space .
it implies that the low - ion transition lines overlap over the two identified clustered absorption line systems @xmath19 the sub - dla and the absorber at @xmath20 shifted by only @xmath21 km s@xmath16 from the sub - dla ( see above ) .
it may thus be possible that some of the components of the detected low - ion metal lines are associated with this second absorber and not with the sub - dla system .
two distinct clumps in the low - ion profiles are not yet observed , but the two high column density absorption line systems are so close that it would not be surprising to find their metal lines blended .
moreover , the observed high - ion transition lines , and , are surprisingly confined to a much narrower velocity interval from @xmath22 to @xmath23 km s@xmath16 ( see fig .
[ j2344-fits ] and table [ j2344-values ] ) .
this may suggest that the detected high - ion lines are associated only with the sub - dla system .
if this is the case , the components of the low - ion transition lines belonging to the sub - dla could be only the ones which are located in the same velocity interval as the high - ion lines , i.e. the components 7 to 11 .
the components 1 to 6 could then be associated with the second absorption line system .
it is difficult to be categorical on the real situation , therefore it would perhaps be more sensible to speak of an interval of values for the column density measurements in this sub - dla system .
in the 12 sub - dla systems and 1 borderline case between the dla and sub - dla systems studied , we obtained ionic column density measurements of low - ion transitions , o@xmath24 , c@xmath25 , si@xmath25 , n@xmath24 , s@xmath25 , mg@xmath25 , al@xmath25 , fe@xmath25 , ni@xmath25 , zn@xmath25 and cr@xmath25 , intermediate - ion transitions , al@xmath26 and fe@xmath26 , and high - ion transitions , c@xmath27 and si@xmath27 . to derive the _
intrinsic _ abundance measurements of the 11 detected elements , it is necessary to take into account both the ionization and dust depletion effects , since we are measuring gas - phase abundances .
depletion is discussed in the wider context of paper ii .
here we analyse the ionization status of these systems .
several authors have investigated the ionization effects in the dla systems ( viegas 1995 ; howk & sembach 1999 ; izotov , schaerer & charbonnel 2001 ; vladilo et al .
2001 ; prochaska et al . 2002 ) .
many of them were motivated by the observations of the intermediate - ion al@xmath26 , a tracer of moderately ionized gas , with the same velocity profiles as the low - ion transitions ( lu et al .
1996b ; prochaska & wolfe 1999 ; prochaska et al .
2001 ) , which suggest that ionization effects may be present in these systems . indeed , to explain the similarity of and other low - ionization species line profiles , howk & sembach ( 1999 ) and izotov & thuan ( 1999 ) proposed that these lines originate in the same ionized region or in a mix of neutral and ionized clouds , and stressed the importance of abundance corrections for ionization effects .
although the different approaches used to deal with the problem of photoionization in dlas have led to slightly different conclusions , it is generally accepted that ionization corrections in dlas are negligible , being below the measurement errors . in the present study
, we are concerned about the ionization corrections for sub - dla , i.e. systems with column densities lower than those of dla systems , between @xmath28 ( ) @xmath29 cm@xmath1 .
the check of ionization effects in these systems is very important , since their relatively low column density implies that some of the gas might be ionized . in six out of the 12 sub - dla systems
studied we detect the @xmath301854,1862 lines , and in four we also detect the @xmath301122 line . while neutral and singly ionized species with ionization potential @xmath31 ev can be present in the inner parts of h@xmath24 regions , the production of fe@xmath26 and al@xmath26 requires photons with @xmath32 and 18.8 ev , respectively ,
which can not so easily penetrate large column densities .
the analysis of these intermediate - ion transition profiles , however , shows that in all of the sub - dlas but one at @xmath33 towards q1101@xmath19264 ( see section [ q1101 ] ) , their velocity structure is almost indistinguishable from that of low - ion transition profiles , while the high - ion and transitions have very different velocity profiles .
this suggests that the fe@xmath26 and al@xmath26 and regions of neutral h in these systems are physically connected .
/al@xmath25 column density ratios versus column densities in sub - dlas ( filled symbols ) and dlas ( open symbols ) for all the systems found in the literature and studied in this paper .
the squares correspond to the data points obtained with the al@xmath25 column density measurements , while the circles and the triangles correspond to the data points computed from the si@xmath25 and fe@xmath25 column density measurements , respectively , according to the deduced linear regression parameters given in table [ correlations ] .
the dashed line represents a possible upper band limit of an increasing dispersion in the @xmath6(al@xmath26)/@xmath9(al@xmath25 ) ratios with lower column densities .
the slope of this upper band limit is equal to the slope of the observed @xmath6(al@xmath26)/@xmath9(al@xmath25 ) versus @xmath6(h@xmath24 ) anti - correlation in dlas.,width=302 ] the @xmath9(al@xmath26)/@xmath9(al@xmath25 ) and @xmath9(fe@xmath26)/@xmath9(fe@xmath25 ) ratios can then be used as indicators of the ionization state .
vladilo et al .
( 2001 ) studied the dla ionization properties from the analysis of the al@xmath26/al@xmath25 ratios in a data sample of 20 dlas , and found a @xmath34(al@xmath26)/@xmath9(al@xmath25 ) versus @xmath6(h@xmath24 ) anti - correlation .
the existence of such an empirical anti - correlation implies that the neutral hydrogen column density is an indirect tracer of the ionization state of the gas .
consequently , if this anti - correlation is further observed in the sub - dla column density range , it would indicate that the sub - dla systems require higher ionization corrections than absorbers in the dla column density range .
therefore , we decided to investigate the extension of the different correlations found by vladilo et al .
( 2001 ) in the dla systems to the column density range of the sub - dla systems . for this purpose
we have collected all the dla and sub - dla systems found in the literature for which measurements of the al@xmath25 and/or al@xmath26 column densities exist and was performed on high resolution spectra obtained at the 8 - 10 m class telescopes .
most of them were derived from the hires / keck spectra .
we have then considered only the systems which show similarities in their al@xmath26 and low - ionization species line profiles , since it is only in this case that the neutral and ionized clouds might be physically related .
finally , we have adopted the values given by vladilo et al .
( 2001 ) for the few cases of al@xmath25 column density measurements that they have revisited due to previously unquoted saturations or contaminations by telluric lines .
the results and the references for the adopted sample of both dlas and sub - dlas found in the literature and analysed in this paper are given in table [ correlations ] .
this table contains the spearman s test correlation coefficients , @xmath35 , the probabilities of null hypothesis ( of no correlation ) , @xmath2 , the slopes and the intercepts obtained for the linear regressions of @xmath6(si@xmath25 ) versus @xmath6(fe@xmath25 ) , @xmath6(al@xmath25 ) versus @xmath6(si@xmath25 ) , @xmath6(al@xmath25 ) versus @xmath6(fe@xmath25 ) , and @xmath6(al@xmath26)/@xmath9(al@xmath25 ) versus @xmath6(h@xmath24 ) .
we confirm the existence of the @xmath6(si@xmath25 ) versus @xmath6(fe@xmath25 ) , @xmath6(al@xmath25 ) versus @xmath6(si@xmath25 ) , and @xmath6(al@xmath25 ) versus @xmath6(fe@xmath25 ) correlations found by vladilo et al .
( 2001 ) in the dla systems ( see table [ correlations ] ) .
these correlations are also satisfied by the sub - dla systems .
we then used the si@xmath25 ( in priority , because the correlation with al@xmath25 has lower slope and intercept errors ) or fe@xmath25 column densities as a proxy for @xmath9(al@xmath25 ) , when no @xmath9(al@xmath25 ) measurement was available in the dla or sub - dla systems . in four sub - dlas and several dlas we took advantage of the correlations si@xmath25 versus al@xmath25 and fe@xmath25 versus al@xmath25 to indirectly estimate @xmath9(al@xmath25 ) in order to increase the number of the @xmath6(al@xmath26)/@xmath9(al@xmath25 ) ratio measurements .
figure [ aliii - alii ] shows the al@xmath26/al@xmath25 column density ratios versus h@xmath24 column densities both for the dla ( open symbols ) and sub - dla ( filled symbols ) systems .
we confirm the @xmath6(al@xmath26)/@xmath9(al@xmath25 ) versus @xmath6(h@xmath24 ) anti - correlation for the enlarged sample of dla systems ( 30 objects instead of 17 used by vladilo et al .
2001 ) at the probability of null hypothesis of @xmath36 ( correlation coefficient of @xmath37 ; see table [ correlations ] ) .
however , we show that at a lower column density regime , this correlation is no longer valid .
the sub - dla systems have al@xmath26/al@xmath25 ratios which are similar to those of the dlas and not higher , as shown in fig .
[ aliii - alii ] with five measured values and seven upper limits ( filled symbols ) . only one system at @xmath38 towards pss j2155 + 1358 with @xmath6 ( ) @xmath39 is really in agreement with the extrapolation of the dla correlation to lower column densities ( filled square in the upper left corner in fig . [ aliii - alii ] ) .
this implies that the neutral hydrogen column density is _ not _ a reliable indicator of the ionization state of the gas in absorption line systems as suggested by vladilo et al .
( 2001 ) , since it does not scale with the al@xmath26/al@xmath25 column density ratio which is at a first approximation an indicator of the ionization level in the gas . on the contrary , figure [ aliii - alii ]
shows that since the sub - dla systems , despite their lower column densities , span approximately the same range of the al@xmath26/al@xmath25 column density ratio values as the dla systems , they are likely to be affected by ionization corrections on average at the same level as the dlas and not higher .
however , we can not claim that all of the absorption line systems with column densities between @xmath40 and @xmath3 cm@xmath1 ( the sub - dla column density definition ) satisfy the same trend , as already shown by the example of the sub - dla system at @xmath41 towards pss j2155 + 1358 .
we can rather interpret the present data as showing an increasing dispersion in the @xmath6(al@xmath26)/@xmath9(al@xmath25 ) ratios with lower h@xmath24 column densities .
the plotted dashed line in figure [ aliii - alii ] with the slope of the @xmath6(al@xmath26)/@xmath9(al@xmath25 ) versus @xmath6(h@xmath24 ) anti - correlation observed in dlas can be considered as _
an upper band limit _ of this increasing dispersion trend .
if this trend is further confirmed by additional measurements , this would suggest that at high @xmath9(h@xmath24 ) @xmath42 cm@xmath1 only low ionization effects are expected , while at lower @xmath9(h@xmath24 ) between @xmath40 and @xmath3 cm@xmath1 one can find systems affected by low as well as high ionization corrections . to investigate the ionization corrections in our sub - dlas ,
we have used the cloudy software package ( version 94.00 , ferland 1997 ) , and we have computed photoionization models assuming ionization equilibrium . in the models we have adopted a plane - parallel geometry for the gas cloud , the solar abundance pattern , and two different ionizing spectra , since the major uncertainty in determining the ionization effects in absorption line systems is the unknown shape of the ionizing spectrum .
the two most likely origins for ionizing uv photons are external background sources and internal stellar sources @xmath19 it is reasonable to expect some star formation in these systems . for the external ionizing source
, we adopted the haardt - madau ( hereafter hm ) ionizing uv spectrum from background quasars with contributions from galaxies ( haardt & madau 1996 ; madau , haardt & rees 1999 ) .
for the stellar ionizing source , we adopted the spectrum of a local ionization which we consider to be a composite stellar spectrum plus a power law ( hereafter ssp ; see dodorico & petitjean 2001 ) . in the former case
the ionizing photons have a harder spectrum . with each of these two ionizing spectra
, we computed a series of cloudy models for each sub - dla system in our sample by varying the column density of the plane - parallel ` cloud ' , the metallicity and the redshift according to the system considered , and by varying the volume density of hydrogen , @xmath43 , through the variation of the ionization parameter , @xmath44 , defined by @xmath45 where @xmath46 is the surface flux of ionizing photons with @xmath47 ryd , and @xmath48 is the intensity of the incident radiation at 1 ryd . in practice , we varied @xmath49 from @xmath50 to 0 , and obtained the theoretical column density predictions for all of the observed ions as a function of the ionization parameter .
we then determined the ionization state of a given sub - dla system by measuring adjacent ions of elements ( e.g. al@xmath26/al@xmath25 , fe@xmath26/fe@xmath25 ) .
indeed , the value of the ionization parameter , @xmath44 , at which the observed column density ratio of adjacent ions matches the predicted ratio , corresponds to the ionization parameter of the system studied .
therefore , the more numerous the measurements of column density ratios of adjacent ions of different elements , the more accurate is the determination of the ionization state of a system , all the more since the reliability of the derived photoionization models depends not only on the assumed hypothesis , but also on the accuracy of the input atomic data for the cloudy code ( e.g. dielectric recombination coefficients ) . following the prochaska et al .
( 2002 ) ionization analysis , we have further computed models giving the predicted abundance of low - ion x@xmath51 over h@xmath24 and fe@xmath25 relative to the intrinsic abundance of these two elements , * @xmath52 } = \log ( n({\rm x_i})/n({\rm h}^0 ) ) - \log ( { \rm x / h})_{\odot}$ ] , * @xmath53 } = \log ( n({\rm x_i})/n({\rm fe}^+ ) ) - \log ( { \rm x / fe})_{\odot}$ ] , as a function of the ionization parameter , @xmath44 . departures of [ x@xmath51/h@xmath24 ] and [ x@xmath51/fe@xmath25 ] from zero indicate that ionization corrections have to be added to the observed absolute and relative elemental abundances , respectively .
when the corrections to be applied are negative ( positive ) , this implies that the measured [ x / h ] or [ x / fe ] are underestimated ( overestimated ) .
the main advantage of these models compared to the models giving the predicted x@xmath51 column densities as a function of @xmath44 is that they allow a direct quantification of the effects of photoionization on the absolute and relative abundances . in the sample of the 12 sub - dla systems and
one borderline case between the dla and sub - dla systems studied here , the column densities range from @xmath6 ( ) @xmath56 to 20.30 cm@xmath1 , and the metallicities spread over @xmath57 dex ( from [ fe / h ] @xmath58 to [ fe / h ] @xmath59 ) .
this large range of input parameters for the cloudy models allows us to highlight some general trends .
indeed , the computed cloudy models specific to each sub - dla system in our sample show that for a given ionization parameter , @xmath44 , the derived [ x@xmath51/h@xmath24 ] and [ x@xmath51/fe@xmath25 ] ionization corrections ( as defined in section [ methodology ] ) increase with lower column densities of systems for all low - ions x@xmath51 and for both the hm or ssp ionizing spectra , while no particular trend is observed with the metallicity of the systems .
in addition , we notice that the ionization corrections are on the average more important in the models with a ssp radiation field than with a hm spectrum for all of the systems studied .
we show in figs .
[ q1223-ionization ] and [ q2116-ionization ] the cloudy models for two sub - dla systems , one at @xmath60 towards q1223 + 1753 with a very low column density @xmath6 ( ) @xmath61 and a high [ fe / h ] @xmath62 , and the second at @xmath63 towards pss j2155 + 1358 with a high column density @xmath6 ( ) @xmath64 and a low [ fe / h ] @xmath65 , respectively .
they illustrate these general trends .
once we have computed the cloudy models , the determination of the ionization state is crucially dependent on the measurements of column density ratios of adjacent ions of elements , and thus on the detections of intermediate - ion transitions . in the sample of 13 systems
studied , we were able to constrain the ionization parameter , @xmath44 , for 10 systems .
however , we obtained for only four systems the @xmath9(al@xmath26)/@xmath9(al@xmath25 ) and/or @xmath9(fe@xmath26)/@xmath9(fe@xmath25 ) measurements , while for the six remaining systems we derived only @xmath9(al@xmath26)/@xmath9(al@xmath25 ) and/or @xmath9(fe@xmath26)/@xmath9(fe@xmath25 ) upper limits .
the analysis of the ionization state of our sub - dla systems is thus severely limited by the fact that _ i ) _
we do not have a measurement of a column density ratio of adjacent ions of an element for all of the sub - dlas , and _
ii ) _ for the majority of sub - dlas
we have only a measurement of the al@xmath26/al@xmath25 ratio or even only an upper limit , and as has already been pointed out by petitjean , rauch & carswell ( 1994 ) , dodorico & petitjean ( 2001 ) and vladilo et al .
( 2001 ) , the al@xmath26/al@xmath25 ratio is not a very reliable indicator of the ionization parameter .
indeed , the recombination coefficient of al@xmath25 is likely overestimated ( nussbaumer & storey 1986 ) , and in this case the predicted al@xmath26/al@xmath25 ratio in the cloudy models is underestimated .
this prevents us from obtaining a very accurate determination of the ionization parameter in our sub - dla systems .
in addition , with our set of data , we are not able to constrain the nature of the radiation field , since we can not clearly differentiate which from the hm or ssp ionizing spectrum best represents the observations . in this context
we choose to extract from our analysis general conclusions on the ionization corrections in the sub - dla systems studied , rather than a discussion on individual systems which is limited by the partial information we have .
several important general remarks can be inferred from our study .
_ first _ , as can be observed in figs
. [ q1223-ionization ] and [ q2116-ionization ] , the ionization corrections to apply to absolute and relative abundances are different ion to ion . indeed , in the case of c@xmath25 , o@xmath24 , s@xmath25 or cr@xmath25 , for example , the ionization corrections [ x@xmath51/fe@xmath25 ] increase smoothly with the ionization parameter , @xmath44 , and remain relatively small for all of the @xmath44 values , while for the si@xmath25 , al@xmath25 and zn@xmath25 ions they already become important at relatively low @xmath44 values .
similarly , for the metallicities , the [ x@xmath51/h@xmath24 ] corrections are not significant for fe@xmath25 up to high @xmath44 values , while they are much higher for zn@xmath25 at high @xmath44 values .
_ secondly , for 9 systems out of 10 for which we were able to constrain the ionization parameter the ionization corrections are lower than 0.2 dex for all of the elements but and _ ( in some cases ; mainly if we assume that the ssp radiation field is the dominant ionizing source in our systems ) down to column densities of @xmath6 ( ) @xmath7 cm@xmath1 ( see figs . [ q1223-ionization ] and [ q2116-ionization ] ) . in the case of al@xmath25 and
zn@xmath25 the ionization corrections may be overestimated , since the al@xmath25 and zn@xmath25 recombination coefficients and ionization cross sections are rather uncertain ( see nussbaumer & storey 1986 and howk , savage & fabian 1999 , respectively ) . only in the sub - dla system at @xmath38 towards pss j2155 + 1358 are the ionization corrections important ( filled square in the upper left corner in fig .
[ aliii - alii ] ) . for three sub - dlas in our sample
we could not constrain the ionization parameter , @xmath44 .
the constructed sample of 12 sub - dla systems and one borderline case between the dla and sub - dla systems , and the homogeneous chemical abundance analysis of each system presented in this paper result in the first sub - dla chemical abundance database ideal for the study of a number of important issues .
indeed , 16 different ions of low- , intermediate- and high - ionization and 11 different elements @xmath19 o , c , si , n , s , mg , al , fe , ni , zn and cr @xmath19 have been detected in the 13 systems studied , showing a variety of elements as diverse as that observed in the dlas .
tables [ n - summary ] a ) and b ) present a summary of the column density measurements of the low - ion transitions and the intermediate-/high - ion transitions , respectively , obtained in the 13 systems analysed .
these column densities were derived by summing the column densities of all of the detected velocity components in each ion as given in tables [ q1101-values]@xmath19[j2344-values ] .
the cases in which we provide only lower limits correspond to the cases where the observed metal lines are saturated , and the cases in which we provide only upper limits correspond to cases where the observed metal lines are blended with ly@xmath2 forest absorption lines .
4 @xmath66 upper limits were derived in the cases where the metal lines are not detected and are identified by an asterisk . in table [ absolute - summary ] we provide a summary of the absolute abundance measurements of the 12 sub - dla systems and one borderline case between the dla and sub - dla systems , defined with respect to the solar meteoritic abundances from grevesse & sauval ( 1998 ) .
no ionization correction has been adopted since , as was shown in section [ ionization ] for all of the systems studied for which we were able to constrain the ionization parameter ( 10 systems out of 13 ) except one at @xmath38 towards pss j2155 + 1358 , the ionization corrections are almost negligible , being at the level of the errors .
indeed , they are lower than 0.2 dex for all of the elements except and ( in some cases , but likely overestimated because the al@xmath25 and zn@xmath25 recombination coefficients are very uncertain ) down to column densities of @xmath6 ( ) @xmath7 cm@xmath1 .
[ abundance - plots ] shows the distribution of iron and silicon absolute abundances , [ fe / h ] and [ si / h ] , as a function of redshift and column density for the 12 sub - dlas in our sample . the studied sub - dlas span a range of redshift from 1.8 to 4.3 and of [ fe / h ] metallicities from @xmath67 to @xmath68 , which is very similar to that observed in dlas .
we observe a trend of increasing metallicity with redshift , but no correlation between metallicities and column densities is visible
. an overall discussion on the chemical abundances of sub - dlas , and a detailed comparison of the chemical , kinematic and clustering properties between the sub - dlas and dlas is given in paper ii .
.,width=321 ] the lower column densities of the sub - dla systems relative to the dla systems give us the opportunity to have access to some elements which absorption lines are usually saturated in dlas , but prevent us from measuring abundances of some other elements whose lines are already weak in dlas . as examples of the former case , one can quote the carbon and oxygen . indeed ,
c is almost not accessible in dlas , since the @xmath301036,1334 lines are heavily saturated .
only very few tentative measurements have been obtained ( lopez et al .
1999,2002 ; levshakov et al . 2002
; songaila & cowie 2002 ) . the o abundance is also very difficult to measure in dlas , since the most commonly detected line at @xmath69 is saturated and other lines are located in the ly@xmath2 forest . on the contrary ,
in the sub - dla systems one can more easily measure the abundances of these two important elements @xmath19 we obtained four o and six c abundance measurements @xmath19 which may provide insight into the understanding of the nature and the chemical evolution history of these systems ( see paper ii ) . on the other hand ,
the useful zn abundance used as an indicator of the dust depletion in dlas is very hard to measure in sub - dlas given the weakness of the lines .
while numerous zn abundance measurements exist in dlas , we obtained only two zn measurements in our sample of sub - dlas .
the findings by proux et al .
( 2003a ) that the ` sub - dla ' systems with column densities between @xmath28 ( ) @xmath29 cm@xmath1 contain a large fraction of the neutral hydrogen gas mass especially at @xmath70 , have motivated us to investigate systematically quasar spectra in the eso uves - vlt archives searching for these absorbers .
we have obtained the following results : * out of 22 high - resolution quasar spectra available to us in july 2001 in the eso uves - vlt archives and suitable for the construction of a sample of sub - dla systems , reduced and analysed with an identical approach , we have built the first sample of 12 sub - dla systems in the redshift interval @xmath5 .
these systems were identified in the quasar spectra both via an automated selection algorithm and visual criteria , like the presence of damping wings in the ly@xmath2 absorption line .
ten of them constitute a sample that we believe to be homogeneous and unbiased . * the detailed metal line analysis of each of the 12 sub - dla systems and one borderline case between the dla and sub - dla systems resulted in the column density measurements of 16 ions of low- , intermediate- and high - ionization @xmath19 , , , , , , , , , , , , , , and .
the detection of adjacent ions of the same element , fe@xmath26/fe@xmath25 and al@xmath26/al@xmath25 , allowed us to estimate the ionization parameters in 10 systems . using the cloudy software package , we computed photoionization models and found that for all of the 10 systems except one at @xmath38 towards pss j2155 + 1358 , the ionization corrections are almost negligible , being at the level of the errors .
indeed , they are lower than 0.2 dex for all of the elements except and ( in some cases , but likely overestimated because the al@xmath25 and zn@xmath25 recombination coefficients are very uncertain ) down to column densities of @xmath6 ( ) @xmath7 cm@xmath1 .
* we have obtained the first sub - dla chemical abundance database which contains the abundance measurements of 11 different elements @xmath19 o , c , si , n , s , mg , al , fe , ni , zn and cr . the lower column densities of sub - dlas allowed us , in particular , to easily measure the o and c abundances , the and lines being usually saturated in dlas .
we obtained o abundance measurements in four systems and c in six systems .
the constructed homogeneous and unbiased sample of the sub - dla systems and the obtained sub - dla chemical abundance database are ideal for the analysis of a number of important issues related both to the neutral gas mass and the metal content evolutions in the universe .
these issues are discussed in paper ii of this series , where we also present a detailed study of the sub - dla kinematic , chemical and clustering properties , and their comparison with the properties of the well studied dla systems .
we plan to extend the current sample of sub - dla systems in order to increase the statistical weight of the obtained results both from the newly accessible data in the eso uves archives and from observations of pre - selected high - redshift quasars , since it is at high redshift @xmath70 that the sub - dlas are expected to contain a large fraction of the neutral hydrogen gas mass ( proux et al .
we are very grateful to paolo molaro and miriam centurin for providing us with the reduced and normalized spectrum of the quasar q0000@xmath192620 .
we particularly thank miriam centurin for her help in the computation of photoionization models .
we express our gratitude to the eso staff in charge of the vlt archive for their efficient work to the benefit of the community .
we thank an anonymous referee for helpful comments .
z . is supported by the swiss national funds and c.p .
is funded by a european marie curie fellowship .
l c c c c c c + + name & @xmath71 & @xmath6 ( ) & @xmath6 ( ) & @xmath6 ( ) & @xmath6 ( ) & @xmath6 ( ) + q1101@xmath19264 ............. & 1.838 & 19.50(0.05 ) & 12.34(0.04 ) & ... & 14.22(0.04 ) &
13.84(0.05 ) + q1223@xmath721753 ............ & 2.557 & 19.32(0.15 ) & 12.32(0.13 ) & 14.31(0.13 ) & 14.44(0.04 ) & 13.99(0.02 ) + q1409@xmath72095 ............. & 2.668 & 19.75(0.10 ) & 11.89(0.05 ) & 13.17(0.12 ) & ... & ... + q1444@xmath72014 ............. & 2.087 & 20.18(0.10 ) & @xmath73@xmath14 & @xmath74@xmath14 & 13.22(0.05 ) & 12.94(0.04 ) + q1451@xmath72123 ............. & 2.255 & 20.30(0.15 ) & 12.70(0.05 ) & ... & ... & ... + q1451@xmath72123 ............ & 3.171 & 19.70(0.15 ) & @xmath75@xmath14 & ... & 13.48(0.13 ) & ... + q1511@xmath72090 ............ & 2.088 & 19.47(0.10 ) & @xmath76@xmath14 & @xmath77@xmath14 & ... & ... + q2059@xmath19360 ............. & 2.507 & 20.21(0.10 ) & @xmath78@xmath14 & ... & ... & ... + q2116@xmath19358 ............. & 1.996 & 20.06(0.10 ) & 13.25(0.09 ) & @xmath79 & @xmath80 & 14.38(0.02 ) + pss j2155@xmath721358 ..... & 3.142 & 19.94(0.10 ) & @xmath75 & ... & ... & ... + pss j2155@xmath721358 ..... & 3.565 & 19.37(0.15 ) & 12.71(0.09 ) & @xmath81 & 14.38(0.10 ) & 14.05(0.04 ) + pss j2155@xmath721358 ..... & 4.212 & 19.61(0.10 ) & ... & ... & ... & 12.93(0.12 ) + pss j2344@xmath720342 ..... & 3.882 & 19.50(0.10 ) & @xmath82@xmath14 & ... & 14.19(0.05 ) & 13.64(0.07 ) + | in this first paper of a series , we report on the use of quasar spectra obtained with the uves high resolution spectrograph and available through the eso vlt archive to build the first homogeneous and unbiased sample of sub - dla systems , absorbers with column densities @xmath0 cm @xmath1 but lower than the classical definition of damped ly@xmath2 systems ( dlas ) @xmath3 cm@xmath1 . a systematic investigation of the properties of these systems and a comparison with those of the dlas ( paper ii of this series ; proux et al .
2003b ) is expected to provide new clues on the association of high column density absorbers with galaxies and on the overall evolution of the neutral hydrogen gas mass and metal content in the universe . in the spectra of 22 quasars which were found suitable for a sub - dla search
, we identified 12 sub - dlas and 1 borderline case between the dla and sub - dla systems in the redshift interval @xmath5 .
we measured the column densities of and of up to 16 additional ions of low- , intermediate- and high - ionization , , , , , , , , , , , , , , , , and .
we further investigated the significance of the ionization corrections in the determination of the chemical abundances from the low - ionization ions in the sub - dla column density range . using the predictions of different ion ratios as a function of the ionization parameter computed with the cloudy software package
, we have estimated that with the exception of one case , the ionization corrections to the abundances of 9 systems for which we were able to constrain the ionization parameter , are lower than 0.2 dex for all of the elements except and down to column densities of @xmath6 ( ) @xmath7 cm@xmath1 .
we finally present the first sub - dla chemical abundance database which contains the abundance measurements of 11 different elements ( o , c , si , n , s , mg , al , fe , ni , zn , and cr ) . we took advantage of the lower column densities in sub - dlas to measure , in particular , the o and c abundances using lines which are normally saturated in dlas .
# 1#2#1#2 [ firstpage ] galaxies : abundances galaxies : high - redshift quasars : absorption lines quasars . |
we study a class of knots generalizing torus knots , which we call _ lissajous - toric _ :
a torus knot is generated by a a circle rotating around an axis and a lissajous - toric knot is generated by a lissajous curve rotating around the axis .
there are several ways of describing them .
we recall the description of the @xmath0-torus knot in @xmath1 endowed with an orthonormal frame @xmath11 ( see for exemple [ cr ] 1.5 ) . if @xmath12 is the circle of radius @xmath13 centered at @xmath14 in the @xmath15 plane , a point travelling along the knot goes @xmath2 times around @xmath12 while @xmath12 is rotated @xmath3 times around the axis @xmath16 .
+ in the case of the _ lissajous - toric knots _ , we replace the vertical circle by a vertical lissajous curve : we take three integers @xmath17 with @xmath18 and a real number @xmath19 , and we define a knot @xmath20 as follows .
consider the curve @xmath21 given in a vertical plane by @xmath22\longrightarrow\mathbb{r}^3\ ] ] @xmath23 and rotate @xmath21 is rotated @xmath3 times around the axis generated by @xmath24 . in cartesian coordinates , we write the knot as @xmath25 we write ( * ) above in cylindrical coordinates : @xmath27 thus @xmath20 is a closed @xmath3-braid which we can write in the @xmath26-cylinder @xmath28 as follows : @xmath29 note the similarity with the @xmath0-torus knot which is written in the @xmath26-sphere or the @xmath26-cylinder as @xmath30 just as billiard curves are equivalent to lissajous knots ( cf .
[ jp ] ) lissajous toric knots are equivalent to billiard curves in a square solid torus , namely a cube where the top and bottom have been identified ; c. lamm introduced them in [ la 1 ] , see also the related [ l - o ] .
such billiard curves are parametrized similarly to lissajous toric knots ; the trigonometric functions are replaced by _ saw - tooth functions _ of the type @xmath31-\frac{1}{2}|.$ ] and @xmath32 $ ] .
@xmath33 & \longrightarrow [ -1,1]^3 \\ t & \mapsto \left ( g(n.t ) , g(p.t+\phi ) , h(q.t ) \right ) \end{array } \right)\ ] ] c. lamm noticed that these billiard curves in a solid torus do not depend on the phase up to mirror transformation and stated that , if @xmath34 and @xmath2 are mutually prime , the knot @xmath4 is ribbon .
+ we first encountered the @xmath20 s in [ s - v ] when we studied the singularities of minimal disks in @xmath35 ; having noticed that their knot types do not depend on the phase @xmath19 up to mirror transformation , we dropped the @xmath19 in the notation .
+ we consider a minimal , i.e. conformal harmonic , map @xmath36 where @xmath37 is the unit disk in @xmath38 , with @xmath39 , i.e. @xmath40 has a branch point at @xmath41 .
if moreover @xmath40 is a topological embedding , we can copy milnor s construction of algebraic knots ( [ mi ] ) and take the intersection of @xmath42 with a small sphere centered at @xmath43 : we obtain a _ minimal knot_. complex curves are a special case of minimal surfaces and the germ @xmath44 yields the @xmath0-torus knot . in [ s - v ]
the knots @xmath20 s came from germs of singularities of the type @xmath45 with @xmath46 in [ s - v ] we called the @xmath4 s _ simple minimal knots _ ; in the present paper we drop the assumption ( [ hypothese riemannienne ] ) and study these knots _ per se _ ; _ lissajous - toric _ is a more appropriate name for the general case . in [ s - v ]
we defined a braid @xmath6 naturally associated to the knot @xmath4 ; we describe it here in much greater detail .
we view @xmath6 as a collection of graphs of @xmath3 functions from @xmath47 $ ] to @xmath48 ; the purpose of the small positive number @xmath49 is to avoid crossing points at the endpoints of the interval .
we prove in [ preuve de la proposition sur les tresses periodiques ] below [ ce qui nous mene a lassumption ] let @xmath50 , @xmath51 , @xmath52 ; then @xmath53 since @xmath54 and @xmath55 are mutually prime and since the knot type does not change if we interchange @xmath34 and @xmath2 , we make the [ assumption on the integers ] the numbers @xmath34 and @xmath2 are mutually prime and @xmath2 is odd . in [ paragraph - description of the simple minimal braid ] , we construct two braids @xmath56 and @xmath57 of the form @xmath58 where the @xmath59 s are the standard generators of the braid group @xmath60 and the exponents @xmath61 of the @xmath59 s appearing in @xmath62 and @xmath63 are given by simple formulae in @xmath17 .
+ + we will state below the main theorem which expresses the braid @xmath6 as a product of the braids @xmath56 , @xmath64 , @xmath57 and @xmath65 as follows : @xmath66 where the @xmath3-braid @xmath67 is also a product of @xmath68 s and @xmath69 s .
we illustrate the main theorem in
[ section sur les exemples ] by going through the examples we gave in [ s - v ] and we prove it in [ the structure of a simple minimal braid : proofs ] .
+ in the rest of the paper , we drop the assumption [ assumption on the integers ] and study the topology of the knot . in [ preliminaires sur le ribbon ] , we prove a theorem stated by lamm [ theoreme sur le ribbon ] if @xmath34 and @xmath2 are mutually prime , the knot @xmath4 is ribbon .
[ corollaire sur les noeuds periodiques ] if @xmath70 , the knot @xmath4 is periodic and its braid is the @xmath9-th power of a braid which closes in a ribbon knot .
[ theoreme sur le four - genus ] if @xmath50 , the four - genus of @xmath4 verifies @xmath71 the right - hand side of ( [ inegalite pour le four genus ] ) is the genus of the @xmath72-torus knot ( cf .
[ k - m ] ) .
the inequality ( [ inegalite pour le four genus ] ) can be strict : for example the knot @xmath73 is @xmath74 which is slice .
there is one case where we know that ( [ inegalite pour le four genus ] ) is an equality : [ four genus ; egalite ] let @xmath17 be positive integers with @xmath18 , @xmath75 and let @xmath76 @xmath77 the knot @xmath4 is represented by a quasipositive braid and its @xmath10-genus is @xmath78 finally , replacing @xmath79 by @xmath80 in the expression of @xmath4 given in [ a description in r3 ] yields if @xmath34 and @xmath2 have different parities ( and thus @xmath3 is odd ) , then @xmath4 is preserved by the involution @xmath81 hence it is positive strongly amphicheiral .
some of the @xmath4 s are actually trivial knots ; in [ la section sur les noeuds triviaux ] show : [ liste de noeuds triviaux ] if @xmath3 and @xmath2 are mutually prime , the knots @xmath82 , @xmath83 , @xmath84 and @xmath85 are trivial .
can we get all the trivial @xmath4 s this way ?
we did computer simulations using the braid software from the liverpool knot group ( [ br ] ) and knotplot ( [ kp ] ) : they told us that in some cases ( the @xmath86 s for example ) the answer is yes but in most cases the answer is no ( see the lists of jones polynomials at the end of the paper ) .
we are grateful to moshe cohen whose stimulating conversation prompted us to embark on this work .
here is an informal description of the contents of the main theorem .
+ + there are @xmath87 values of @xmath79 in @xmath88 $ ] ( we call them _ crossing values _ ) above which two or more of the @xmath3 graphs forming @xmath6 meet ( at _ crossing points _ ) and the data of all these crossing points make up the braid ( see figure [ premiere image de tresse ] ) ; above each of the crossing values @xmath79 s , the generators of the braid group @xmath60 describing the corresponding crossing points are all even ( i.e. of the form @xmath89 ) or all odd ( i.e. of the form @xmath90 ) .
+ + the set of crossing points of @xmath6 above a crossing value @xmath79 can be represented by one of the braids : @xmath56 , @xmath64 , @xmath57 or @xmath65 which were introduced in formula ( [ la premiere fois quon voit alpha et beta ] ) ; thus @xmath6 is a product of the @xmath56 s and @xmath57 s and of their inverses .
+ we order the @xmath87 crossing values @xmath91 .
going from @xmath92 to @xmath93 changes @xmath68 into @xmath69 or vice - versa .
a formula gives us the exponent @xmath94 or @xmath95 of the @xmath56 or @xmath57 above a given crossing point @xmath92 in terms only of @xmath17 and @xmath96 .
+ finally we notice that , if we have an @xmath56 ( resp .
@xmath64 , @xmath57 , @xmath65 ) for @xmath92 ( with @xmath97 ) , we have a @xmath64 ( resp .
@xmath56 , @xmath65 , @xmath57 ) for @xmath98 : this explains the presence of @xmath67 and @xmath99 in the product ( [ la premiere fois quon voit b comme un commutateur ] ) .
[ structure generale ] let @xmath100 be three integers such that @xmath2 is odd and @xmath101 ; and let @xmath102 , @xmath103 two integers such that @xmath104 for @xmath105 , we let @xmath106}\ ] ] where @xmath107 $ ] denotes the integral part and we define @xmath108 for @xmath109 , @xmath110 , we let @xmath111}\ ] ] up to mirror transformation , the knot @xmath4 is represented by the braid @xmath112 the @xmath96-th factor in this expression corresponds to the @xmath96-th crossing value @xmath92 .
notice that the arithmetic formulae ( [ definition depsilon avec le signe de conway ] ) and ( [ definition de lambda avec le signe de conway ] ) can be written in terms of the conway sign : ( [ co ] ) [ definition du signe de conway ] if @xmath113 and @xmath114 are two integers , @xmath114 is said to be _ positive _ ( resp .
_ negative _ ) modulo @xmath113 if @xmath114 is congruent to an integer inside @xmath115 ( resp .
@xmath116 ) .
in this section we go through the examples featured in [ s - v ] and we write their braid using the terminology of the main theorem . + we define three permutations of the crossing values @xmath117 and their corresponding action on the blocks @xmath62 and @xmath63 : @xmath118 since it is clear in each case of the following list what the @xmath17 are , we dropped the indices @xmath17 . *
@xmath119 : square knot @xmath120 @xmath121 * @xmath122 : trivial knot @xmath123 * @xmath124 : figure eight knot @xmath125 * @xmath126 : @xmath127 @xmath128 note that this knot verifies the assumptions of theorem [ four genus ; egalite ] .
* @xmath129 : @xmath74 @xmath130 * @xmath131 : @xmath132 @xmath133 * @xmath134 : @xmath135 @xmath136 * @xmath137 : @xmath138 @xmath139 * @xmath140 : @xmath141 @xmath142 * @xmath143 : @xmath144 @xmath145
we recall some facts from [ s - v ] . we endow @xmath1 with coordinates @xmath146 : the braid is the collections of the graphs in @xmath1 of the functions @xmath147 for @xmath148 : @xmath149\longrightarrow\mathbb{r}^2\ ] ] @xmath150 [ image de la tresse 357 ] we now prove proposition [ ce qui nous mene a lassumption ] ( stated in [ subsection : contents of the paper ] ) .
we divide the interval @xmath47 $ ] into @xmath9 intervals @xmath151,\ \ \ \ \ \ \ n=0, ... ,d-1.\ ] ]
after a change of variables @xmath152 , we see that the braid above an interval @xmath153 consists in the collection of graphs of the functions @xmath154\longrightarrow \mathbb{r}^2 $ ] @xmath155 since @xmath156 , the map @xmath157 induces a permutation of @xmath158 ; hence the piece of @xmath6 above @xmath153 is the collection of graphs above @xmath47 $ ] of the functions @xmath159 i.e. it is the braid @xmath160 , representing the knot @xmath161 ; this proves proposition [ ce qui nous mene a lassumption ] .
+ the _ braid shadow _ is the projection of the braid onto the first two components @xmath162 of @xmath48 i.e. the collection of the graphs of the @xmath163 s .
+ a _ crossing point _
@xmath164 of the braid is the data of two different integers , @xmath96 , @xmath165 with @xmath166 and a number @xmath167 $ ] called a _ crossing value _ such that @xmath168 there is a total of @xmath169 crossing points , as in the case of the @xmath0 torus knot ( where @xmath170 ) . + a straightforward computation ( cf .
[ s - v ] ) shows that , for a crossing point @xmath164 between the @xmath96-th and @xmath165-th strands of @xmath6 , the corresponding crossing value @xmath79 verifies for some integer @xmath113 @xmath171 the sign @xmath172 of a crossing point @xmath164 is and @xmath95 ] @xmath174 in [ s - v ] , we computed this @xmath172 as : @xmath175}(-1)^{[p\frac{k - l}{n}]}(-1)^{[q\frac{k - l}{n}]}\ ] ] where @xmath176 $ ] denotes the integral part .
let @xmath79 be a crossing value of @xmath6 .
we look for the @xmath177 s such that @xmath162 is a crossing point of the braid shadow .
+ we derive from ( [ equ : valeur de t ] ) the existence of at least one ordered pair of integers @xmath178 such that @xmath179 + ( with @xmath180 ) .
there can be several @xmath181 s verifying ( [ les m et s pour une crossing value ] ) for the same crossing value @xmath79 ; however @xmath182 which implies that there are at most two possible @xmath181 s ( lemma [ nombre de m et s ] below ) .
we will see later that for @xmath79 and @xmath181 given , a crossing point @xmath162 of the braid shadow above @xmath79 will be given by the data of @xmath183 .
[ nombre de m et s ] if @xmath79 is a crossing value , one of the following two cases occurs : * * 1st case*. there is exactly one ordered pair @xmath181 , @xmath184 verifying ( [ les m et s pour une crossing value ] ) ; we denote it @xmath185 and @xmath186 is either @xmath187 , @xmath188 or @xmath3 . * * 2nd case*. there exist exactly two @xmath181 s satisfying ( [ les m et s pour une crossing value ] ) with @xmath184 ; we denote them @xmath189 and @xmath190 .
we let @xmath191 be the ordered pair such that @xmath192 is the smallest @xmath193 for the @xmath194 s verifying ( [ les m et s pour une crossing value ] ) and ( [ inegalite pour s et m ] ) .
if @xmath195 and @xmath196 both verify ( [ les m et s pour une crossing value ] ) for the same @xmath79 , we have @xmath197 since @xmath2 and @xmath3 are mutually prime , it follows that , for some integer @xmath198 , @xmath199 since @xmath200 is the smallest one , it verifies @xmath201 if we are in the 1st case of the lemma [ nombre de m et s ] , i.e. a single @xmath202 , we derive from ( [ difference entre les s et les m ] ) that @xmath203 does not verify ( [ inegalite pour s et m ] ) , i.e. @xmath204 putting together ( [ plus petit que n ] ) and ( [ s en dehors de lintervalle ] ) , we get @xmath205 which concludes the proof of the 1st case . +
the 2nd case is clear : since @xmath206 , @xmath203 is the only other integer @xmath207 in @xmath208 $ ] which can appear in ( [ les m et s pour une crossing value ] ) ; the corresponding @xmath113 is @xmath209 .
lemma [ nombre de m et s ] told us which @xmath193 s and @xmath113 occur for crossing points @xmath162 above a crossing value @xmath79 : we now find the @xmath210 s such that @xmath211 and derive the @xmath212 s corresponding to the @xmath162 s .
let @xmath213 be a crossing point ; we denote by @xmath214 the corresponding generator subscript , i.e. @xmath164 is represented by @xmath215 or @xmath216 .
[ les valeurs de y ] let @xmath79 be a crossing value of @xmath6 . 1 .
the point @xmath213 is a crossing point of the braid shadow if and only if @xmath217 where @xmath218 is any integer in @xmath219 $ ] of the same parity as @xmath186 2 . to determine @xmath220
, we do the euclidean division of @xmath221 by @xmath222 @xmath223 with @xmath224 , @xmath225 . 1 . if @xmath226 is even , @xmath227 2 .
if @xmath226 is odd , @xmath228 * proof of 1 * : we treat separately the two cases of lemma [ nombre de m et s ] . 1 . * 1st case * : a single ordered pair @xmath181 .
+ let @xmath210 such that @xmath229 and assume that @xmath230 ; note that @xmath183 has the parity of @xmath186 .
1 . if @xmath231 , the smallest possible value for @xmath165 is @xmath41 and @xmath232 runs through all integers @xmath218 , @xmath233 with the parity of @xmath231 .
2 . if @xmath234 ( resp .
@xmath235 ) , the smallest value for @xmath165 is @xmath41 ( resp .
@xmath13 ) and @xmath236 runs through the integers in @xmath237 $ ] with the parity of @xmath187 or @xmath3 ; since @xmath188 has parity opposite to @xmath3 and @xmath187 , we can actually assume @xmath218 in @xmath238 $ ] .
+ to derive ( [ valeur de y ] ) , we plug ( [ les m et s pour une crossing value ] ) into @xmath239 2 . *
2nd case * : two ordered pairs : @xmath185 and @xmath240 .
1 . we first consider the @xmath210 s such that @xmath230 and @xmath241 .
as above , @xmath232 runs through the integers @xmath218 with the parity of @xmath242 and such that @xmath243 2 .
if @xmath244 , we look at @xmath245 s with @xmath230 : @xmath246 @xmath247 @xmath248 moreover every integer in @xmath249 $ ] with the parity of @xmath250 is a legitimate @xmath218 , i.e. there exist @xmath210 in @xmath158 with @xmath183 and @xmath244 ; for example @xmath251 . + + using ( [ lexpression de y ] ) , we derive the @xmath177-coordinate of the crossing point : @xmath252 where @xmath253 ; if @xmath218 verifies ( [ intervalle possible pour delta ] ) , then @xmath254 since @xmath218 has the parity of @xmath250 , @xmath255 has the parity of @xmath256 .
+ putting together the intervals ( [ premiere inegalite sur frakd ] ) and ( [ intervalle pour d tilda ] ) concludes the proof of the 2nd case .
* proof of 2*. we derive from ( [ euclidean division ] ) that @xmath257 so 2 ( a ) of lemma [ les valeurs de y ] follows from the fact that the function @xmath258 is decreasing on @xmath259 : @xmath260 and 2 ( b ) of lemma [ les valeurs de y ] follows from @xmath261 to see how many crossing points @xmath162 occur above @xmath79 , i.e. how many values ( [ valeur de y ] ) takes for a given @xmath79 , we notice the following . * if @xmath262 and @xmath263 , then @xmath264 . * if @xmath262 and @xmath265 , then @xmath266 .
so the sets @xmath267 and @xmath268 are identical .
we derive from lemma [ les valeurs de y ] [ differentes parites des generateurs de tresses ] let @xmath79 be a crossing value .
the indices @xmath220 s for the crossing points @xmath164 s above @xmath79 are all the @xmath269 s in @xmath158 and with 1 .
the parity of @xmath270 and @xmath200 if @xmath226 is even 2 .
the parity of @xmath271 and @xmath272 if @xmath226 is odd .
we now compute the sign of the crossing points described in lemma [ les valeurs de y ] .
the sign of the crossing point corresponding to the @xmath162 appearing in ( [ valeur de y ] ) is given by @xmath273 where @xmath274 is the parity of the integer part : @xmath275}\ ] ] we recall ( lemma [ nombre de m et s ] ) that @xmath229 or @xmath276 .
+ if @xmath229 and @xmath183 , then ( [ signe de y en fonction de u ] ) is just the formula ( [ formula pour le signe ] ) for the sign of a crossing point .
so we assume that @xmath244 : we have seen above that @xmath177 is given by ( [ formule pour y pour m plus q ; deux ] ) with @xmath277 and we write ( [ formula pour le signe ] ) for the sign of the crossing point @xmath278 @xmath279 we write the last two factors of ( [ signe de y en fonction de u ] ) in terms of @xmath220 : [ les deux derniers facteurs du signe ] we let @xmath280 be a crossing point of @xmath6 with @xmath226 , @xmath218 and @xmath281 as in 2 ) of lemma [ les valeurs de y ] .
then @xmath282 it follows from ( [ euclidean division ] ) that @xmath283 we recall ( [ definition des a et b ] ) , namely @xmath284 , hence @xmath285 ; putting this together with the euclidean division in ( [ euclidean division ] ) , we have @xmath286 thus @xmath287 now @xmath288 * if @xmath226 is even , then @xmath289 and the lemma is proved . * if @xmath226 is odd , we use the fact that @xmath103 is odd to write @xmath290 it follows from lemma [ les deux derniers facteurs du signe ] that the sign @xmath172 given in ( [ signe de y en fonction de u ] ) of a crossing point @xmath164 of crossing value @xmath79 is ( @xmath291 has been defined in ( [ definition depsilon avec le signe de conway ] ) above ) * @xmath292 if @xmath34 is even * @xmath293 if @xmath34 is odd .
we recall ( see the formulae ( [ deuxieme apparition de alpha et beta ] ) in the main theorem ) that the @xmath61-exponent of a @xmath59 in @xmath56 or @xmath57 is @xmath294 , hence [ exposant de alpha et beta ] the @xmath61-exponent of the @xmath56 or @xmath57 corresponding to a crossing value @xmath79 in @xmath6 is 1 .
@xmath295 if @xmath34 is even 2 .
@xmath296 if @xmath34 is odd .
the formulae in lemma [ exposant de alpha et beta ] depend on @xmath226 , where @xmath79 goes through the @xmath87 crossing values .
if @xmath79 is the @xmath297-th crossing value , for @xmath298 , we want to have @xmath226 directly as an expression in @xmath297 so we number the crossing values @xmath299 and for any integer @xmath96 , with @xmath300 , we let @xmath301 if @xmath92 and @xmath93 are two consecutive crossing values , we derive from ( [ les m et s pour une crossing value ] ) @xmath302\geq \frac{1}{2q}\ ] ] since there are @xmath87 crossing values in @xmath47 $ ] , ( [ crossing values qui se suivent ] ) is an equality and we have @xmath303 we note in passing that , if we plug ( [ relation entre les s et m pour les valeurs qui se suivent ] ) into proposition [ differentes parites des generateurs de tresses ] , we get the confirmation of the obvious fact the crossing points above @xmath92 and @xmath93 are represented by @xmath304 s with @xmath269 s of opposite parities .
we now confront ( [ relation entre les s et m pour les valeurs qui se suivent ] ) with @xmath284 and derive the existence of an integer @xmath305 such that @xmath306 thus , for any @xmath96 , there exists an integer @xmath307 such that @xmath308 @xmath309 define @xmath310 by @xmath311 we see ( [ formula pour le signe ] ) that @xmath310 is a critical phase , i.e. a phase for which the knot @xmath312 is singular .
so we pick a phase @xmath313 where @xmath314 is a very small positive number . using ( [ une expression
pour mk ] ) , we rewrite @xmath315}\ ] ] it follows from lemma [ exposant de alpha et beta ] and equations ( [ une expression pour mk ] ) , ( [ expression pour s avec xi ] ) that the exponent of the @xmath56 or @xmath57 at the @xmath96-th crossing value is @xmath316}\mbox{\ \ if\ } p\mbox{\ is odd}\\ \\
( -1)^{\big[\frac{2ap}{q}k+\xi\big]}\mbox{\ \ if\ } p\mbox{\ is even } \end{array } \right.\ ] ] since we are working up to mirror transformation , we assume @xmath317 we now conclude : the expression @xmath318 $ ] is equal to @xmath13 for @xmath319 and equal to @xmath320 for the other @xmath96 s . going back to the statement of the main theorem , this gives us the exponent for the @xmath96-th crossing values with @xmath321 or @xmath322 .
we settle the case of the @xmath96 s with @xmath323 by noticing that @xmath324
a _ ribbon knot _ in @xmath325 bounds a disk in @xmath325 with only ribbon singularities ; equivalently it bounds an embedded disk in @xmath326 with does not have local maxima for the distance to the origin of @xmath326 . thus a ribbon knot is slice , i.e. its @xmath10-genus is zero ; the long - standing slice - ribbon conjecture asks if the converse is true .
[ theoreme sur le ribbon ] has been stated by lamm and also follows from his more general construction of ribbon symmetric unions ( [ la 2 ] , [ k - t ] ) .
his proof is fairly allusive so we felt it would be useful to give a more detailed proof .
+ we recall the well - known fact : [ noeud symetrique est ribbon ] let @xmath327 be a knot in @xmath1 which is symmetric with respect to a plane @xmath164 in @xmath1 ; then it is ribbon .
we endow @xmath1 with the frame @xmath11 and assume that @xmath164 is defined by the equation @xmath328 . by genericity arguments , we assume 1 .
@xmath327 meets @xmath164 at a finite number of points 2 . outside of @xmath164 ,
@xmath327 is never tangent to the direction of @xmath329 .
since @xmath327 has one component , it meets @xmath164 at exactly two points .
+ we let @xmath330 ( resp .
@xmath331 ) be the intersection of @xmath327 with the half - space of @xmath1 defined by @xmath332 ( resp .
@xmath333 ) : @xmath330 and @xmath331 are both diffeomorphic to a closed interval
. + letting @xmath334 be the symmetry in @xmath1 with respect to @xmath164 , we let @xmath335\times k_+\longrightarrow\mathbb{r}^3\ ] ] @xmath336 the self - intersections of @xmath337 are given by the data of @xmath338 such that @xmath339 we denote by @xmath340 , @xmath341 the coordinates of @xmath342 .
since @xmath343 , ( [ egalite ] ) implies that @xmath344 thus the line segments @xmath345 and @xmath346 are both included in the line which is defined by the equations @xmath347 .
moreover , one of them is included in the other one and we have a ribbon singularity .
the main theorem tells us that , if @xmath34 and @xmath2 are mutually prime , with @xmath2 odd , @xmath6 is as in fig .
[ premier dessin ] .
+ [ h ] [ deuxieme dessin ] if @xmath3 is the number of strands , there are @xmath188 half - twist tangles connecting @xmath348 and @xmath349 ; we replace them by @xmath188 tangles and get the @xmath3-component link @xmath350 of fig .
[ deuxieme dessin ] which is symmetric w.r.t . a plane .
proposition [ noeud symetrique est ribbon ] tells us that @xmath350 bounds @xmath3 ribbon disks @xmath351 , ... , @xmath352 ; and
the same arguments show us that two of these disks only have ribbon - type intersection .
+ we now connect each @xmath353 to @xmath354 by a half - twisted band bounded by the half - twist tangle of fig .
[ premier dessin ] .
the resulting surface is a topological disk with only ribbon singularities .
we use an idea by brandenbursky and kedra ( [ b - k ] ) .
if @xmath355 is a @xmath3-braid , we denote by @xmath356 the link obtained by closing the braid @xmath355 . if @xmath357 and @xmath358 are @xmath3-two braids , [ b - k ] constructed a cobordism of euler characteristic @xmath359 between the closure of the product @xmath360 and the disjoint union of the closures @xmath361 .
+ letting @xmath362 , @xmath363 , we recall that @xmath364 applying [ b - k ] s result @xmath9 times , we derive a cobordism in @xmath326 of euler characteristic @xmath365 between @xmath366 and @xmath367 .
since @xmath368 is a ribbon knot ( theorem [ theoreme sur le ribbon ] ) , it bounds an embedded disk in @xmath326 .
thus @xmath366 bounds a surface of euler characteristic @xmath369 we recover the formula ( [ inegalite pour le four genus ] ) for the genus and theorem [ theoreme sur le four - genus ] is proved .
lee rudolph ( see [ ru ] for details ) defines a braid @xmath370 to be _ quasipositive _ if it is a product of conjugates @xmath371 of positive braid generators , i.e. @xmath372 [ rudolph]([ru ] ) if @xmath373 is a closed quasipositive braid written as in ( [ quasipositive ] ) closing in a knot @xmath374 , its four - genus verifies @xmath375 it is easy to check that under the assumptions of proposition [ four genus ; egalite ] , the exponents of all the @xmath376 s and @xmath377 s appearing respectively in @xmath56 and @xmath57 are all of the same sign . since we are working up to mirror symmetry
, we can assume all these exponents to be equal to @xmath13 ; thus @xmath378 hence @xmath379 is a quasipositive braid ; theorem [ rudolph ] tells us that @xmath380 and proposition [ four genus ; egalite ] follows .
we set @xmath381 [ forme
de la tresse triviale ] @xmath382 the main thing to note about this formula is that all the positive generators are on one side and all the negative generators are on the other side .
we use the main theorem .
it assumes that @xmath2 is odd but in the present case , if @xmath2 is even , @xmath3 has to be odd , hence @xmath383 is odd and we switch @xmath2 and @xmath383 to apply the theorem .
we compute @xmath384 thus @xmath385 and @xmath386 .
we conclude by noticing that @xmath387 we construct a trivial pure braid @xmath388 ; we will show that @xmath389 is the product of a power of @xmath388 and of a piece of @xmath390 .
if @xmath391 is even , we let @xmath392 if @xmath393 is odd , we let @xmath394 in both cases , we check that the corresponding permutation between the endpoints of the braid is the identity , thus @xmath395 is a pure braid . + + to prove that it is a trivial braid , we discuss when one strand of @xmath395 is above another one ; so let us fix some terminology .
+ we number the strands of @xmath388 : the @xmath396-th strand , @xmath397 is the strand starting at the @xmath398-th point on the left ( the points being counted from top to bottom ) .
+ we say that the @xmath396-th strand is above the @xmath96-th strand if , wherever there is a crossing point between these two strands , the @xmath396-th strand is above the @xmath96-th strand . as an exemple , in fig .
[ second dessin ] , the red strand is above all the other strands . [ au dessus de toutes les suivantes ] [ brins au dessus ] if @xmath399 are two integers with @xmath400 , the @xmath396-th strand of @xmath395 is above the @xmath96-th strand .
thus @xmath395 closes in @xmath3 unlinked trivial links , i.e. @xmath401 . the figure [ second dessin ] illustrates the lemma .
we describe the strands of @xmath395 in the braid shadow , i.e. their projection to the @xmath402 .
we endow the plane with a coordinate @xmath403 such that the @xmath396-th strand starts at @xmath404 and ends at @xmath405 .
the upper left point has coordinates @xmath406 ( in figure [ second dessin ] it is the starting point of the red strand ) .
+ we say that a strand is _ ascending _ , denoted @xmath407 ( resp .
_ descending _ , denoted @xmath408 ) if it has a @xmath94 ( resp .
@xmath95 ) slope .
it is _ horizontal _ , denoted @xmath409 , when the slope is @xmath41 .
+ + we describe here the @xmath96-strands for @xmath96 odd ( the case of an even @xmath96 is similar ) : it goes up and down as follows + 1 .
@xmath408 from @xmath410 to @xmath411 2 .
@xmath409 from @xmath411 to @xmath412 3 .
@xmath407 from @xmath413 to @xmath414 4 .
@xmath409 from @xmath414 to @xmath415 5 .
@xmath408 from @xmath415 to @xmath416 assume now that the @xmath96-th strand is above the @xmath396-th strand at a crossing point @xmath417 .
assuming that @xmath396 is odd ( the even case is similar ) , one of the following two cases occurs 1 .
@xmath418 and the @xmath96-th ( resp .
@xmath396-th ) strand is @xmath408 ( resp .
@xmath407 ) .
then @xmath419 is as 1 . above and
@xmath420 is as 3 . 2 .
@xmath421 and the @xmath96-th ( resp .
@xmath396-th ) strand @xmath407 ( resp .
@xmath408 ) .
then @xmath419 is as 3 . above and
@xmath420 is as 5 . in both cases
it is easy to check that @xmath422 .
we conclude the proof of the proposition in the case when @xmath3 is even ; the odd case is similar .
we derive from lemma [ au dessus de toutes les suivantes ] that for an @xmath423 , @xmath424
@xmath425 thus , if @xmath355 is the remainder of the division of @xmath426 by @xmath96 , we have @xmath427 the braid ( [ forme finale de la tresse triviale pour n pair ] ) is a piece of the braid @xmath428 where the @xmath269-th strand is above the @xmath396-th strands , for @xmath429 .
thus the same is true for ( [ forme finale de la tresse triviale pour n pair ] ) which closes therefore in a trivial knot .
this follows from the main theorem .
we can also prove it directly by computing the crossing points and their sign : we see that every @xmath212 appears once and only once in the braid @xmath431 and so braid represents a trivial knot .
we have now seen two cases where @xmath4 is trivial .
we know that @xmath4 and @xmath432 are isotopic ; and @xmath433 and @xmath434 ( resp .
@xmath435 ) are isotopic ( resp .
mirror image of one another ) .
thus we can get more examples of trivial knots , e.g. @xmath436 and @xmath437 .
we give a better proof of the following fact from [ s - v ] : without loss of generality , we assume @xmath442 . + if there is no critical phase ( i.e. a phase for which the knot is singular ) between @xmath438 and @xmath439 , the two knots are isotopic . + in [ s - v ] we showed that the difference between two critical phases is of the form for two integers @xmath444 . +
thus it is enough to prove that , for a given @xmath445 , and integers @xmath113 and @xmath114 , the knots @xmath446 and @xmath447 are the same or mirror images of one another .
+ consider the parametrization of @xmath20 given in ( [ definition de la tresse ] ) ; we change its variable by setting @xmath448 and we rewrite the expression in ( [ definition de la tresse ] ) @xmath449 @xmath450 thus , if @xmath113 and @xmath114 have the same ( resp .
opposite ) parities , the two knots are isotopic ( resp . mirror image of one another ) .
br the liverpool knot group _ br9z.p _ , https://www.liverpool.ac.uk/ su14/knotprogs.html m. brandenbursky , j. kedra _ concordance group and stable commutator length in braid groups _ , arxiv:1402.3191 ( 2014 ) , to appear in algebraic & geometric topology ( 2015 ) . g. burde , h. zieschang , _ knots _ , de gruyter studies in math . ,
5 , walter de gruyter & co , new york , 2003 .
conway assisted by fung , f. y. c. _ the sensual ( quadratic ) form _ maa ( 1997 ) .
p. cromwell _ knots and links _ cambridge university press , 2014 .
j. franks , r. williams _ braids and the jones - conway polynomial _ , trans .
( 1987 ) 97 - 108 l.h . kauffman _ on knots
_ , pup , 1987 .
http://www.knotplot.com/ v.f.r .
jones , j.h .
lissajous knots and billiard knots _ knot theory , banach center publications , 42 , inst . of maths , polish acad . of sciences ,
warsaw 1998 p. kronheimer , t. mrowka , _ the genus of embedded surfaces in the projective plane _ , math .
letters 1 , 797 - 808 ( 1994 ) s. kinoshita , h. terasaka _ on unions of knots _ , osaka j. math . 9 ( 1957 ) , 131 - 153 . c. lamm , d. obermeyer _ billiard knots in a cylinder _ j. knot theory and its ramifications 8(3 ) ( 1999 ) vol . 353 - 366 . c. lamm _ deformation of cylinder knots _ fourth chapter of ph.d .
zylinder - knoten und symmetrische vereinigungen _
bonner mathematische schriften 321 ( 1999 ) , http://arxiv.org/pdf/1210.6639 ( 2012 ) c. lamm _ symmetric unions and ribbons knots _ , osaka j. math .
37 ( 2000 ) , 537 - 550 x. mo , r. osserman , _ on the gauss map and total curvature of complete minimal surfaces and an extension of fujimoto s theorem _ , jour .
geom . 31 ( 1990 ) , 343 - 355 .
m. micallef , b. white , _ the structure of branch points in minimal surfaces and in pseudoholomorphic curves _ , annals of maths , 139 , ( 1994 ) , 35 - 85 j. milnor , _ singular points of complex hypersurfaces _ , ann . of math .
studies , pup ( 1968 ) .
d. rolfsen , _ knots and links _ , publish or perish , houston , 1990 .
l. rudolph , _ quasipositivity as an obstruction to sliceness _ , bull . of the ams , 29 ( 1 ) ( 1993 )
pages 51 - 59 m. soret , m. ville , _ singularity knots of minimal surfaces in @xmath35 _ , journal of knot theory and its ramifications , 20 ( 04 ) 2011 , 513 - 546 . | a point in the @xmath0-torus knot in @xmath1 goes @xmath2 times along a vertical circle while this circle rotates @xmath3 times around the vertical axis . in the lissajous - toric knot @xmath4
, the point goes along a vertical lissajous curve ( parametrized by @xmath5 while this curve rotates @xmath3 times around the vertical axis .
such a knot has a natural braid representation @xmath6 which we investigate here . if @xmath7 , @xmath4 is ribbon ; if @xmath8 , @xmath6 is the @xmath9-th power of a braid which closes in a ribbon knot .
we give an upper bound for the @xmath10-genus of @xmath4 in the spirit of the genus of torus knots ; we also give examples of @xmath4 s which are trivial knots . |
this work is part of the research program of the foundation for fundamental research on matter ( fom ) , which is financially supported by the netherlands organisation for scientific research ( nwo ) .
s.y.t.v.d.m . acknowledges support from nwo via a vidi and a top grant , and from the european research council via a starting grant .
we thank sean gordon for assistance during a research visit and david parker for fruitful discussions .
the expert technical support by leander gerritsen , chris berkhout , peter claus , niek janssen and andr van roij is gratefully acknowledged . | we present ultrahigh - resolution measurements of state - to - state inelastic differential cross sections for no - ne and no - ar collisions , obtained by combining the stark deceleration and velocity map imaging techniques .
we show that for counterpropagating crossed beam geometries , the effect of the velocity spreads of the reagent beams on the angular resolution of the images is minimized .
futhermore , the counterpropagating geometry results in images that are symmetric with respect to the relative velocity vector .
this allows for the use of inverse abel transformation methods that enhance the resolution further .
state - resolved diffraction oscillations in the differential cross sections are measured with an angular resolution approaching 0.3@xmath0 .
distinct structures observed in the cross sections gauge the quality of recent _ ab initio _ potential energy surfaces for no - rare gas atom collisions with unprecedented precision .
the development of experimental methods to study molecular collisions with the highest possible detail has been a quest in molecular collision research since it was established in the 1950s . in recent years , velocity map imaging ( vmi ) has revolutionized our ability to study molecular collision processes in crossed beam experiments @xcite .
the power of the technique derives from the ability to record a two - dimensional image of scattered molecules that directly reflects their recoil velocities and thus the differential cross section ( dcs ) of the scattering process . in the last decade
, vmi has been implemented in many scattering experiments and has made experiments possible that would have been inconceivable only a few years ago @xcite .
precise measurements of dcss provide unique and sensitive tests for the underlying potential energy surfaces ( pess ) that govern the dynamics of interacting molecules . to validate the ever more sophisticated and accurate theoretical descriptions of scattering processes , it is essential to obtain ever higher resolutions in the experimental scattering images .
the inherent resolution of vmi in principle allows for dcs measurements that probe pess with spectroscopic resolution .
the image resolution that is typically obtained in crossed beam experiments , however , is limited by the velocity spreads of both beams involved .
the overlap between the signals from collision partners with different velocities can significantly blur the images , thereby obscuring detailed structures that may be present in the dcs . in the last decade
, several techniques have been developed to obtain complete control over molecules in beams .
in particular , methods such as stark and zeeman deceleration allow for the production of packets of molecules with high density , tunable velocity , near - zero velocity spread and almost perfect quantum state purity @xcite .
clearly , these tamed molecular beams are ideally suited for precise molecular scattering experiments , in particular when combined with state - of - the art detection techniques such as vmi .
recently , we have reported the first crossed beam experiment in which a stark decelerator was combined with vmi , indeed resulting in scattering images with much higher resolution @xcite . yet ,
thus far only one of the reagent beams has been velocity - controlled using a stark decelerator .
the image resolution is then limited by the relatively large velocity spread of the collision partner which is produced via conventional beam methods .
although reduction of this spread using mechanical velocity filters appears feasible , it will be difficult to reach the narrow distribution that is readily achieved for the stark - decelerated reagent beam . here
, we present a method to minimize the contribution of the collision partner s velocity spread to the angular resolution of the scattering images . exploiting the favorable kinematics present
when counterpropagating beams are used , we obtain record - high angular resolutions and thereby probe dcss with unprecedented accuracy .
this is illustrated by fully resolving diffraction oscillations in state - to - state dcss for no - ne and no - ar collisions .
for no - ar , we observe distinct features in the diffraction pattern that appear extremely sensitive to the details of the pess .
these features are used to gauge the quality of two recent pess for this benchmark system .
the influence of the crossed beam geometry on the kinematics and resolution of the experiment is illustrated in figure [ fig : kinematics ] .
for the sake of the argument , newton diagrams are shown for the scattering of two beams containing particles of equal mass and equal pre - collision mean velocity .
beam crossing angles of 90@xmath0 and 180@xmath1 are assumed in panels _ a _ and _ b _ , respectively .
we approximate the kinematics present in our crossed beam experiments ( a stark - decelerated beam with mean velocity @xmath2 crosses with a conventional beam with mean velocity @xmath3 ) by assuming that only the conventional beam has a non - zero velocity spread . to model a structured dcs
, we consider a hypothetical dcs that exhibits a series of block functions , indicated by colored stripes , that are spaced with an angular interval of 12.5@xmath1 .
newton diagrams are shown for scattering with the mean velocity ( blue ) , and the two outermost values from the beam distribution ( red and green ) .
it is seen that for scattering angles around forward scattering ( @xmath4 ) where the most detailed structures in dcss are typically found the _ velocity _ spread of the collision partner does not contribute to the image blurring when a counterpropagating beam geometry is used .
the angular image resolution is then exclusively determined by the _
angular _ spread of the beam , and high angular image resolutions can be achieved by simply collimating the beam .
this sharply contrasts the situation for a 90@xmath1 crossing angle . here , the velocity spread of the collision partner results in both radial _ and _ angular blurring of the image ; angular image resolutions can only be significantly improved by reducing the velocity spread of the collision partner .
the counterpropagating geometry has the additional advantage that scattering images are inherently cylindrically symmetric with respect to the mean relative velocity vector of the colliding beams ( see figure [ fig : kinematics ] ) .
this allows for the use of inverse abel transformation methods to retrieve the angular scattering distributions that underly the three - dimensional newton spheres .
this transformation effectively cancels the additional blurring of images that occurs when these newton spheres are projected onto the detector plane @xcite .
again , this sharply contrasts the situation for a 90@xmath1 crossing angle . here , the inherent angular and radial resolution is different in every part of the image , and symmetry with respect to the mean relative velocity is broken .
further asymmetry in image intensity arises that is related to a detection bias for low laboratory recoil velocities which are preferentially found on one side of the image @xcite .
together , these asymmetries have thus far prevented the use of abel inversion methods in crossed beam scattering experiments .
methods to improve image resolutions by recording only a central slice of the newton sphere have been developed @xcite , but in crossed beam experiments one can not always afford the inherent reduction in signal intensity .
either way , complex forward iterative methods are required to disentangle instrument dependent detection probabilities from the desired scattering intensity distribution .
the use of counterpropagating beam geometries in crossed beam experiments is rather unconventional .
henning meyer successfully used counterpropagating pulsed beam scattering techniques in the early 90 s to facilitate measurements of angular resolved state specific product distributions using time - of - flight detection of the scattered molecules @xcite , but the technique has hardly been used ever since .
we use a crossed beam scattering apparatus in which a packet of no ( @xmath5 ) radicals with a mean velocity of 350 m / s , a velocity spread of 1.2 m / s ( 1@xmath6 ) , and an angular spread of 0.04@xmath1 ( 1@xmath6 ) is produced by passing a beam of no through a 2.6 m long stark decelerator @xcite .
the packet of no collides with a beam of ne or ar atoms at a beam intersection angle of 180@xmath1 , resulting in a relatively high collision energy of 720 @xmath7 and 725 @xmath7 for no - ne and no - ar , respectively .
the rare gas atom beam is well collimated by placing a 1 mm wide slit between the 2 mm diameter skimmer and the interaction region .
scattered molecules are detected state - selectively using a two - color laser ionization scheme and a standard vmi geometry @xcite .
the two laser beams are directed parallel to the detector plane , and cross each other at right angles .
the implementation of a counterpropagating beam geometry in our experiment is aided by the use of reagent beam pulses with a narrow temporal distribution , such that the pulses only spatially overlap within a well defined volume . in the interaction region ,
the packet of no radicals and the rare gas atom beam has a temporal width of 25 @xmath8s and 15 @xmath8s , respectively .
the experiment is synchronized such that the leading edges of both beams overlap for only about 5 @xmath8s before both detection lasers are fired .
figure [ fig : raw - data](_a _ ) and ( _ b _ ) depict the experimental scattering images for inelastic collisions of no ( @xmath9 ) with ne and ar that excite the no radicals to the @xmath10 and the @xmath11 state , respectively . a clear oscillatory structure
originating from the quantum mechanical nature of molecules that leads to diffraction of matter waves during molecular collisions is observed for both scattering systems .
the @xmath12 channel for no - ne has a relatively large integral cross section of 2.7 @xmath13 , and the well - separated diffraction oscillations in this system could be observed in our previous experiments using a @xmath14 beam crossing angle @xcite .
comparison with the present results shows that when a counterpropagating geometry is used , a superior angular resolution is obtained , even though the collision energy is higher resulting in a smaller peak - to - peak separation of the diffraction peaks .
the @xmath15 channel for no - ar has an integral cross section of only 0.77 @xmath13 , and despite various efforts we have been unable to resolve oscillations in the dcs for this system in previous experiments .
the channel displays one of the richest diffraction structures that we have identified thus far , with a peak - to - peak separation between individual diffraction peaks of 1.6@xmath1 ( _ vide infra _ ) .
the clearly resolved oscillations observed here for this system are a testimony of the unprecedented angular resolution afforded by the counterpropagating geometry .
as expected , the images are ( almost ) symmetric with respect to the relative velocity vector . within our experimental geometry ,
doppler and collision induced alignment effects cause a small dependence of the detection probability on the angle between the recoiling molecule and the propagation direction of the excitation laser . however
, this asymmetry in intensity does not affect the inherent radial and angular resolution on either side of the image , and can easily be accounted for from the known geometry of the experiment .
we find that the images are symmetric after correction for the doppler effect ; no indications were found that collision induced alignment plays a significant role for the inelastic channels studied here . a detection bias for low laboratory recoil velocities results in an additional image intensity correction factor that is equal on either side of the image .
after correction , the symmetric images allow for the use of inverse abel transformation methods to enhance the resolution further . using the basex suite of programs @xcite , images as shown in figure [ fig : abel - data]_a _ and _
b _ are obtained for the two scattering systems .
the oscillatory structure is retrieved with a much higher resolution , as is also clear from the three - dimensional representation of the data shown in panels _ c _ and _ d_. figure [ fig : curves ] shows the angular intensity distributions ( red curves ) that are inferred from the experimental image intensities within a narrow annulus around the rims of the images .
the dcss at the mean collision energies of the experiment , that are predicted by high level _ ab initio _ quantum mechanical close - coupling ( qm cc ) calculations that use the most recent no - ne and no - ar pess presently available @xcite , are shown as well ( black dashed curves ) .
the experimentally obtained scattering distributions are fully consistent with the dcss predicted by theory . in particular the angles at
which diffraction peaks are found are in excellent agreement with the theoretical predictions . by definition ,
the theoretical dcss presented in figure [ fig : curves ] predict the angular distributions that would be measured in the ` perfect ' experiment in which no beam spreads or other factors that limit experimental resolution are present .
it is seen that in our experiments , beam spreads and other experimental factors reduce the visibility of the diffraction structure by about a factor 2 - 3 only .
we estimate the angular resolution of the experiment by convoluting the dcss predicted by the qm cc calculations with a gaussian distribution of variable width .
best agreement with the experimental distributions is found for widths ( @xmath16 ) of 0.45@xmath1 and 0.35@xmath1 for no - ne and no - ar , respectively .
the observed diffraction structure is an excellent proxy for the quality of computed pess .
for the @xmath17 channel of no - ar , there is one specific peak observed at @xmath18 , as indicated by the green arrow , that appears to be extremely sensitive to the exact details of the pes .
this is illustrated in figure [ fig : small - peak]_a _ that shows the scattering intensity around forward scattering for no - ar on an enlarged scale .
the experimentally obtained distribution is compared with the distributions that are expected from two different no - ar pess : one developed by cybulski _
et al .
_ in 2012 and one by alexander from 1999 @xcite .
both potentials are of high quality , but minor differences in calculation methodology and fitting procedures result in two slightly different pess , as illustrated in panel _
b_. it is seen that the existence of this peak in the dcs is predicted by only one of the pess , indicating that the level of precision reached in our experiments is sufficient to discriminate between the most state - of - the - art theoretical methods presently available .
surprisingly , our experiments show best agreement with the oldest pes used . in this letter
, we have shown that in crossed beam scattering experiments , the ability to resolve structures in dcss can be significantly enhanced using a beam crossing angle of 180@xmath1 . for scattering angles around forward scattering , the contribution of the velocity spread of the beam(s ) to the angular resolution
is effectively eliminated .
the beneficial kinematics afforded by the counterpropagating geometry allows for the use of inverse abel transformation methods , facilitating the measurement of dcss with high sensitivity and angular resolution .
this allows for the observation of detailed structures in the dcss that provide unique and stringent tests to gauge the quality of pess and quantum scattering calculations .
our approach appears particularly beneficial in scattering experiments that employ stark - decelerated beams , but crossed beam experiments using two conventional beams may also benefit from a 180@xmath1 crossing angle .
it is noted that the arguments outlined here for a counterpropagating geometry also hold for merged beam scattering , i.e. , experiments that use a beam crossing angle of 0@xmath1 @xcite .
this offers prospects for achieving optimal image resolutions in experiments that aim for low collision energies ; an interesting approach to probe dcss at collision energies where the scattering is dominated by quantum scattering resonances . |
the two - dimensional toda lattice is an important integrable system with many aspects .
it is an example of a nonabelian chern - simons theory in classical field theory ( see @xcite ) , and in differential geometry it can be interpreted as the equation for primitive harmonic maps taking values in a compact flag manifold ( @xcite,@xcite ) .
such maps are closely related to harmonic maps into symmetric spaces .
these in turn have many geometrical interpretations , e.g. surfaces in @xmath1 of constant mean curvature ( see @xcite ) , or special lagrangian cones in @xmath2 ( @xcite,@xcite ) .
we shall be concerned with the two - dimensional periodic toda lattice with opposite sign .
this is the system @xmath3 where @xmath4 is some open subset of @xmath5 .
we assume that @xmath6 for all @xmath7 and @xmath8 .
the first manifestation of the integrability of this system is its zero curvature formulation : @xmath9 where @xmath10 and @xmath11 in other words , ( [ ost ] ) is the compatibility condition @xmath12 $ ] for the linear system @xmath13 the fact that @xmath14 involve only @xmath15 leads to holomorphic data for solutions of ( [ ost ] ) .
we recall from section 4 of @xcite that the holomorphic data is a matrix of the form @xmath16 where each @xmath17 is a holomorphic function . for radial solutions of ( [ ost ] )
, i.e. when @xmath18 depends only on the real variable @xmath19 , ( [ * ] ) reduces to @xmath20 this can be regarded as the equation for an isomonodromic deformation , which is another perhaps more famous manifestation of integrability , and a well known approach to studying equations of painlev type ( see @xcite,@xcite and also @xcite ) .
namely , if we impose the ( euler - type ) homogeneity condition @xmath21 we obtain @xmath22 which is a meromorphic o.d.e . in @xmath23 with poles of order two at @xmath24 and @xmath25 .
writing @xmath26 , we obtain @xmath27 the compatibility condition of the linear system @xmath28 is the radial version @xmath29 $ ] ) of ( [ ost ] ) . from the connection point of view
, the homogeneity assumption ( [ ho ] ) extends the flat connection @xmath30 to a flat connection @xmath31 , where @xmath32 .
it is well known that in this situation the @xmath33-system ( [ muequation ] ) is isomonodromic , i.e. its monodromy data is independent of @xmath34 .
conversely , starting from ( [ muequation ] ) , one may seek its isomonodromic deformations ( see @xcite ) , and this leads back to ( [ xequation ] ) .
the monodromy data of ( [ muequation ] ) consists of formal monodromy and stokes matrices at each of the two poles , as well as a connection matrix relating them .
locally , solutions of the radial version of ( [ ost ] ) correspond to such data .
however this kind of data is in general difficult to compute explicitly , and global properties of the corresponding solution are difficult to read off
. we shall impose the following anti - symmetry condition @xmath35 for some @xmath36 .
for @xmath37 ( equivalently @xmath38 ) , this means that @xmath39 for all @xmath40 ; in this case ( [ ost ] ) and ( [ as ] ) is the system studied by cecotti and vafa in @xcite,@xcite,@xcite . in general we call ( [ ost ] ) and ( [ as ] ) together with the radial assumption the _
tt*-toda equations .
_ in this article we shall identify all solutions of the tt*-toda equations which are smooth on the maximal domain @xmath41 .
they constitute a @xmath42-dimensional family . we shall give simple and explicit parametrizations of this family in terms of asymptotic data , holomorphic data , and monodromy data .
in section [ interp ] we explain the important geometrical and physical motivation behind this description . to end this section
, we mention that equations ( [ * ] ) and ( [ as ] ) possess the following symmetries , which will play an important role in our computation of the monodromy data .
they depend on the three automorphisms @xmath43 of @xmath44 which are defined by @xmath45 where @xmath46 , @xmath47 , and @xmath48 namely , it is easily verified that the connection form @xmath49 has the following properties .
_ cyclic symmetry : _ @xmath50 _ anti - symmetry : _ @xmath51 _ reality : _
@xmath52 the cyclic property expresses the shapes of the matrices in ( [ * ] ) , the anti - symmetry property is just condition ( [ as ] ) , and the reality property corresponds to the fact that every @xmath53 is real .
the differential geometric or lie - theoretic meaning of the three symmetries , and the proof of the following result , can be found in section 2 of our previous article @xcite . from any solution of the system ( [ ost ] ) and ( [ as ] ) we obtain a harmonic map @xmath54 . more generally , it was shown in @xcite ( see also @xcite ) that the tt * equations are locally the equations for ( pluri)harmonic maps into @xmath55 , together with a homogeneity condition ( which amounts to the radial condition in our situation ) .
this is the basis for the geometric interpretation of the tt * equations , as @xmath55 is the space of inner products on @xmath56 .
a solution of the tt * equations can thus be regarded as a ( trivial ) vector bundle of rank @xmath57 on @xmath4 equipped with a metric satisfying various natural conditions .
in fact it is a ( trivial ) harmonic bundle in the sense of @xcite with an additional real structure and homogeneity property . such bundles ( not necessarily trivial ) have been well studied .
they generalize variations of polarized hodge structures , and they appear in @xcite,@xcite,@xcite in connection with semi - infinite variations of hodge structures , pure polarised terp structures , or noncommutative variations of hodge structures .
a special role is played by those bundles which arise from geometry , e.g. from variations of polarized hodge structures on specific manifolds .
the relation with the discussion of the previous section is as follows .
the magical solutions of the tt * equations predicted by cecotti and vafa are expected to be \(1 ) globally defined on @xmath58 , \(2 ) characterized by initial / asymptotic conditions , \(3 ) and have integral stokes data .
we shall refer to such solutions as field - theoretic .
many of them appear to be of algebro - geometric origin , i.e. the corresponding field theories are constructed from unfoldings of singularities ( landau - ginzburg models ) or from quantum cohomology ( nonlinear sigma models ) .
cecotti and vafa predicted the existence of the global solutions on the basis of the physical interpretation of the monodromy data and holomorphic data : the stokes data counts solitons at the infra - red point @xmath59 and the holomorphic data encodes chiral charges at the ultra - violet point @xmath60 .
these are fixed points for the renormalization group flow , and it is this flow that is governed by the tt * equations . for the physical solutions the data at @xmath60 and @xmath59 should have integrality properties .
thus , it has been predicted that the tt * equations admit certain globally smooth solutions with extremely rich geometrical meaning .
our results confirm this prediction in the case where @xmath61 reduce to two independent functions ( in this situation @xmath62 , @xmath63 , or @xmath64 ) .
first , in this case , we describe all solutions of the tt*-toda equations satisfying property ( 1 ) .
the fact that such solutions are in one to one correspondence with asymptotic data at @xmath60 can be proved by using a monotone iteration scheme based on @xcite .
this technique is well known for nonlinear scalar p.d.e . , but novel for systems .
then , by computing the monodromy data and holomorphic data explicitly , we shall verify properties ( 2 ) , ( 3 ) .
the solutions with integer stokes data will be discussed in detail in @xcite ) .
we remark that the holomorphic data is familiar to differential geometers as the generalized weierstrass representation , or dpw representation , of a harmonic map .
the link between the p.d.e .
solution and the holomorphic data is given by the iwasawa factorization @xmath65 of a certain loop group valued function @xmath66 ( with @xmath67 ) .
this is explained in detail in section 4 of @xcite .
it is here that the crucial difference between the toda equations and the tt*-toda equations can be seen . for the toda equations , it is easy to construct solutions with properties ( 1 ) and ( 2 ) , because the relevant lie group is the compact group @xmath68 , and in this case the iwasawa factorization @xmath65 holds on the entire domain of @xmath66 .
in contrast , for the tt*-toda equations we have the noncompact group @xmath69 , and the iwasawa factorization @xmath65 holds ( in general ) only on some open subset @xmath70 of the domain @xmath4 of @xmath66 .
proving that @xmath71 is equivalent to proving that the solution of the p.d.e .
has no singularities , and this is what we shall do .
the same factorization problem arises in the construction of other types of harmonic maps with important differential geometric interpretations ( see , for example , @xcite ) .
as few examples are known where global existence can be proved in the noncompact case , our examples are of interest in this wider context . a detailed statement of results
is given in section [ results ] .
in section [ pdetheorem ] we prove the basic existence theorem ( theorem a ) , and in section [ stokes ] we compute the monodromy data of the solutions ( theorem b ) .
all methods used in this article are low - tech , and our proofs are essentially self - contained . we hope that this will contribute to a better understanding of the tt * equations , which have been treated until now only by indirect methods .
this article can be read independently of @xcite , although our proof of theorem a will make reference to @xcite in order to avoid repetition . in @xcite
we constructed a family of globally smooth solutions of the tt*-toda equations , and we identified a finite number of field - theoretic solutions amongst them . in this article
we complete the picture by constructing _
all _ globally smooth solutions , and we describe _ all _ field - theoretic solutions .
we do this only in the case where @xmath61 are equivalent to two unknown functions ; the general case will be treated in a subsequent article . in @xcite
we give an independent proof of the existence of solutions , from the isomonodromy point of view .
acknowledgements : the first author was partially supported by a grant from the jsps , and the second author by nsf grant dms-1001777 .
both are grateful to taida institute for mathematical sciences for financial support and hospitality during their visits in 2011 , when much of this work was done .
the authors thank claus hertling for his comments on an earlier version of the paper , and also takuro mochizuki for his interest in this project .
mochizuki has given a proof of our main results from the viewpoint of the hitchin - kobayashi correspondence in his preprint harmonic bundles and toda lattices with opposite sign , arxiv : 1301.1718 .
for @xmath72 , we consider the system @xmath73 for @xmath74 .
this is equivalent to ( [ ost ] ) and ( [ as ] ) in the ten cases where @xmath61 reduce to two unknown functions ( and in such cases we have @xmath75 ) .
this data is given in table [ tableofcases ] .
for example , in the case labelled 4a , condition ( [ as ] ) says that @xmath76 and @xmath77 , so the tt*-toda equations reduce to ( [ pde ] ) with @xmath78 if we put @xmath79 , @xmath80 . evidently the cases @xmath81 and @xmath82 correspond under the transformation @xmath83 , so there are in fact just three distinct p.d.e
. systems . .
[ cols="^,^,^,^,^,^,^",options="header " , ] the other entries of table [ integralsolutions ] are discussed in @xcite . for example ( an unfolding of ) the @xmath84 singularity appears as @xmath85 or @xmath86 in the first column , and the @xmath87 singularity appears as @xmath88 in column two or @xmath89 in column three .
these are interior points of the region .
in this section we prove theorem a. for the tt*-toda equations , i.e. @xmath75 , this amounts to solving a certain riemann - hilbert problem , and we intend to discuss these aspects elsewhere . here
, however , we give an elementary approach using the monotone iteration method for solving nonlinear elliptic partial differential equations , and this method works for arbitrary @xmath72 . although the method is well known for scalar p.d.e .
, it rarely applies to systems , and it applies to our system ( [ pde ] ) only through a fortuitous combination of circumstances .
the proof of theorem a in the case @xmath78 and @xmath90 was given in detail in @xcite . in section [ guli1 ]
we extend that proof to the case @xmath91 and @xmath90 . in section
[ guli1extended ] we shall extend the proof further to the case @xmath92 . in section [ allab ]
we explain how the case @xmath91 implies the general case @xmath72 . for notational convenience in this section )
; in that notation we have @xmath93 . ] ( and ease of comparison with section 3 of @xcite ) , we restate the equations as @xmath94 for @xmath95 ( where @xmath72 ) .
our objective ( theorem a ) is to prove that , for any @xmath96 in the region @xmath97 there is a unique solution @xmath98 such that @xmath99 conditions ( [ inequalities ] ) are explained in remark 3.2 ( i),(iii ) of @xcite .
we emphasize that only solutions which are smooth on @xmath58 are to be discussed .
statements such as @xmath100 mean @xmath101 for all @xmath102 .
we do not assume in advance that @xmath53 depends only on @xmath103 ; however , this property follows from the uniqueness of the solution .
the case @xmath105 is analogous to the case @xmath106 , so we just treat the latter .
we shall explain how to extend the proof for @xmath78 in @xcite to the case @xmath91 ; this will also serve as a review of the method of @xcite .
the method has three main steps : \(a ) existence of a supersolution @xmath107 and a subsolution @xmath108 ( this means showing that any solution @xmath98 of ( [ guli1pde ] ) satisfies @xmath109 ) .
\(b ) existence of a maximal solution @xmath98 ( i.e. a solution @xmath98 exists , and if @xmath110 is any other solution , then @xmath111 ) .
\(c ) uniqueness of the solution .
the proof is carried out first for the case where @xmath112 , @xmath113 , then extended to @xmath114 , @xmath115 by a limiting argument . _ proof of ( a ) : _ the fact that @xmath116 follows easily from the maximum principle ( see proposition 3.3 and the following page of @xcite ) .
as in @xcite we have to construct lower bounds for any solution @xmath98 of ( [ guli1pde ] ) . for this ,
let @xmath117 be the solutions of @xmath118 on @xmath58 , where all solutions tend to @xmath24 as @xmath119 .
the existence and uniqueness of @xmath117 can be proved by standard p.d.e .
methods and the maximum principle , under the assumption ( [ inequalities ] ) .
furthermore we can prove that \(i ) @xmath120 \(ii ) @xmath121 \(iii ) @xmath122 and @xmath123 on @xmath58 .
for these proofs we refer to lemma 3.5 , lemma 3.6 and proposition 3.7 of @xcite _ proof of ( b ) : _ we obtain smooth functions @xmath124 ( @xmath125 ) on @xmath58 such that @xmath126 in the following way . first , when @xmath96 lies in the smaller region @xmath127 , we may take @xmath128 and define @xmath129 inductively from @xmath130 by solving the linear elliptic system @xmath131 where @xmath132 ; @xmath133 where @xmath134 .
the subtracted terms on each side are chosen to make @xmath135 and @xmath136 .
it is a key property of the tt*-toda equations that @xmath137 and @xmath135 ( see remark 3.9 ( i ) of @xcite ) .
this monotonicity allows us to establish property ( [ decreasing ] ) .
the sequence @xmath124 converges to a smooth solution @xmath53 on @xmath58 . to prove that it is a maximal solution , the maximum principle is used . to construct ( maximal ) solutions for the range @xmath138 , @xmath139 , @xmath140 we make use of the solutions just constructed .
let @xmath141 be such a solution with @xmath142 as @xmath143 and @xmath144 as @xmath145 .
we choose @xmath146 so that @xmath147 , @xmath148 and @xmath149
. then we take @xmath150 as the first step in the iteration given by the modified linear system @xmath151 where @xmath152 ; @xmath153 where @xmath154 . again
the sequence @xmath124 converges to a maximal smooth solution @xmath53 on @xmath58 .
_ proof of ( c ) : _ by integration
we obtain @xmath155 thus @xmath156 holds for any solution @xmath98 .
suppose @xmath157 is a solution , and @xmath98 is the maximal solution obtained in ( b ) .
then we have @xmath158 since both satisfy ( [ sum ] ) , we must have @xmath159 and @xmath160 .
this completes the proof of ( c ) .
let us consider the system ( [ guli1pde ] ) with @xmath91 where @xmath162 satisfy ( [ inequalities ] ) and also @xmath163 , @xmath164 ( the case @xmath165 , @xmath166 is similar ) .
in particular @xmath167 and @xmath168 . we assume first that @xmath140 ; the case @xmath169 will follow as in section [ guli1 ] .
thus we have the system @xmath170 with boundary conditions @xmath171 and @xmath172 as @xmath145 .
we divide the proof into steps ( a ) , ( b ) , ( c ) as above .
\(a ) let @xmath173 .
the point @xmath174 lies in the region of applicability of section [ guli1 ] , so we have a unique solution @xmath175 such that @xmath176 similarly , for any @xmath177 , there is a unique solution @xmath178 corresponding to @xmath179 , i.e. such that @xmath180 moreover , from section [ guli1 ] , we know that @xmath181 . we claim that any solution @xmath98 of ( [ guli1pdeab1 ] ) must satisfy @xmath182 this will establish lower and upper bounds
. we shall prove first that @xmath183 .
let us suppose that @xmath184 for some @xmath185 .
then the boundary conditions ( in particular @xmath186 ) imply that @xmath187 for some @xmath188 .
hence the maximum principle gives @xmath189 from which it follows that @xmath190 , which implies that @xmath191 .
similarly , the boundary conditions ( in particular @xmath192 ) imply that @xmath193 for some @xmath194 .
therefore @xmath195 applying the maximum principle to the second equation , we have @xmath196 hence @xmath197 as we know @xmath198 .
we obtain @xmath199 which contradicts ( [ 1 ] ) .
this completes the proof that @xmath183 .
the other three inequalities in the claim can be proved in a similar way .
\(b ) next we apply the monotone scheme to prove that there exists a maximal solution , making use of @xmath175 and @xmath178 as supersolution and subsolution , respectively .
we consider the linear equations @xmath200 subject to the boundary conditions @xmath201 and @xmath202 as @xmath145 , where @xmath203 note that @xmath204 and @xmath205 are bounded at @xmath24 ( without singularity ) and @xmath206 there .
we may take @xmath207 close to @xmath208 , so that @xmath209 . thus equation ( [ 2 ] ) is solvable for @xmath129 if @xmath130 is given .
we obtain @xmath130 for @xmath125 by starting with @xmath210 .
note that if @xmath211 satisfies @xmath212 then @xmath213 similarly @xmath214 we shall use this to show that @xmath215 for @xmath125 .
we begin with the case @xmath216 .
first we shall show that @xmath217 .
since @xmath218 , we just have to show that @xmath219 .
from ( [ 2 ] ) we have @xmath220 hence @xmath221 and similarly @xmath222 since @xmath223 and @xmath224 , the maximum principle implies that @xmath225 .
a similar proof shows that @xmath226 .
next we shall show that @xmath227 .
since @xmath228 , properties ( [ 3a])-([3b ] ) give @xmath229 applying this in the equation for @xmath230 , we have @xmath231 so @xmath232 since @xmath233 and @xmath234 , the maximum principle implies that @xmath235 .
similarly we can show that @xmath236 .
this completes the proof of ( [ induction ] ) for @xmath216 . by the same argument it is not difficult to prove ( [ induction ] ) for any @xmath237 .
since @xmath124 is bounded and monotonically increasing with respect to @xmath238 , it converges to some @xmath53 in @xmath239 which evidently satisfies ( [ guli1pdeab1 ] ) .
this completes the proof of existence .
next we claim that the above solution @xmath98 is minimal in the sense that @xmath240 for any other solution @xmath110 with the same @xmath241 .
to prove that @xmath242 , we use the fact already established that @xmath243 .
if we assume @xmath245 . we shall use this to prove that @xmath246 .
we have @xmath247 because @xmath248 is continuous up to @xmath60 .
since @xmath249 as @xmath250 , the maximum principle gives @xmath251 if @xmath252 .
this contradiction shows that @xmath253 . since @xmath254 holds for @xmath216 , by induction it holds for all @xmath255 .
similarly we can prove that @xmath256 for all @xmath255 . by taking the limit @xmath257
we obtain @xmath258 for @xmath259 .
\(c ) the proof of uniqueness of the solution of ( [ guli1pdeab1 ] ) is the same as that in part ( c ) of section [ guli1 ] .
it remains to extend the result for the case @xmath140 to the case @xmath169 .
let us take a sequence @xmath260 in the interior of the region such that @xmath261 , for example @xmath262 with @xmath263 . by the maximum principle ,
the corresponding solution @xmath130 of ( [ 2 ] ) is monotone in @xmath238 and satisfies @xmath264 where @xmath265 , @xmath266 are the solutions of ( [ guli1pdeab1 ] ) corresponding to @xmath267 and @xmath268 . in the limit @xmath257 , @xmath130 converges to a solution of of ( [ guli1pdeab1 ] ) corresponding to @xmath96 .
furthermore , this solution is minimal .
uniqueness of this solution follows by integrating ( [ guli1pdeab1 ] ) .
for these solutions , we note that the weaker asymptotic property @xmath269 holds at @xmath24 .
it suffices to discuss the case @xmath270 and @xmath106 .
the other cases may be dealt with as in sections [ guli1 ] , [ guli1extended ] .
let @xmath178 be the solution to @xmath271 and @xmath110 the solution to @xmath272 with the same @xmath162 .
note that @xmath273 .
these solutions exist by section [ guli1 ] .
next , we claim that @xmath274 if the first inequality does not hold , then @xmath275 .
let @xmath276 we may assume that @xmath277 .
then the maximum principle implies that @xmath278 note that @xmath279 , which implies that @xmath280 .
let @xmath281 by the maximum principle again , we have @xmath282 which is a contradiction
. therefore @xmath283 .
similarly we can prove that @xmath284 .
the claim is proved .
now , @xmath178 is a subsolution of ( [ guli1pde ] ) , because @xmath285 , and @xmath110 is a supersolution of ( [ guli1pde ] ) . since @xmath286 ,
the monotone scheme produces a maximal solution of ( [ guli1pde ] ) .
uniqueness of the solution to ( [ guli1pde ] ) follows by integration as before .
this completes the proof of existence and uniqueness of solutions of ( [ guli1pde ] ) under condition ( [ inequalities ] ) .
given any radial solution of ( [ ost ] ) and ( [ as ] ) on a neighbourhood @xmath287 of a point @xmath288 , we have the corresponding monodromy data of @xmath289 this o.d.e .
has poles of order two at @xmath24 and @xmath25 , so the data consists of collections of stokes matrices @xmath290 , @xmath291 at the poles ( relating solutions on different stokes sectors ) and a connection matrix @xmath292 ( which relates solutions near @xmath24 with solutions near @xmath25 ) . a priori this monodromy data depends on @xmath293 , but it is in fact independent of @xmath185 ; it is a conserved quantity . for convenience
we introduce the new variable @xmath294 and rewrite the o.d.e . as follows : @xmath295 in this section we shall compute the stokes data at @xmath296 , using the approach of @xcite , for the ten cases listed in table [ tableofcases ] . we shall see that each solution referred to in theorem a corresponds to a single stokes matrix at @xmath25 , indeed to just two entries @xmath297 of this stokes matrix . in other words ,
the rest of the monodromy data is uniquely determined by @xmath297 . for each of the ten cases
we shall derive an explicit formula for @xmath297 in terms of the asymptotic data @xmath298 .
. then ( [ 4ode ] ) has the form @xmath300 near @xmath301 . for this o.d.e .
we shall review the formal solutions , the stokes phenomenon ( i.e. the relation between formal solutions and holomorphic solutions ) , then give the definition and computation of the stokes matrices . using this
we give the proof of theorem b. the simplicity of this calculation depends on the fact that the stokes matrices have a very special form as a consequence of the three symmetries of the equation . in order to apply section 1.4 of @xcite ,
we are required to diagonalize the leading term of the coefficient matrix . to do this
we observe that @xmath302 where @xmath303 and @xmath304 where @xmath305 ( the columns of @xmath306 are the eigenvectors of @xmath307 with eigenvalues @xmath308 , where @xmath309 ) .
this gives @xmath310 by proposition 1.1 of @xcite , it follows that there exists a unique formal solution of ( [ 4ode ] ) of the form @xmath311 where @xmath312 .
substitution in the o.d.e .
shows that @xmath313 .
we shall not need this loop group reality condition explicitly ( later we shall use it implicitly , in the proof of proposition [ reduction ] ) .
however we shall need the following more elementary ( and easily verified ) property , which we take as the reality condition from now on : we give the proof of the cyclic symmetry formula ; the other two are similar . from the cyclic symmetry of @xmath327 we see that @xmath328 is also a formal solution , hence must be @xmath329 for some constant matrix @xmath330 . to find @xmath330 we examine @xmath331 since @xmath332 and @xmath333
, we obtain @xmath334 thus , @xmath335 is a formal solution of the same type as @xmath336 , so it must be equal to @xmath336 .
this completes the proof ( and shows that @xmath337 ) .
the stokes phenomenon depends on the fact that ( [ 4ode ] ) admits a holomorphic solution @xmath338 on some region of the form @xmath339 with asymptotic expansion @xmath320 whenever @xmath340 is sufficiently small , i.e. _ if the sector is sufficiently narrow . _
( this will be made precise in a moment . ) on the other hand , it is easy to see that @xmath338 when it exists is unique _ if the sector is sufficiently wide .
_ namely , if @xmath341 are two such solutions , then @xmath342 for some constant @xmath343 , and we have @xmath344 the desired conclusion @xmath345 would follow if , for any @xmath346 , there exists a path @xmath347 in the sector such that @xmath348 since @xmath349 , a sufficient condition is that the path satisfies @xmath350 . if the sector contains a ray @xmath351 with @xmath352 , then @xmath353 can be made positive or negative within the sector , so we have both @xmath354 and @xmath355 . such a ray
is called a stokes ray .
stokes rays are said to have the same type if their arguments differ by @xmath356 .
we conclude that @xmath345 if the sector contains _ at least _
one stokes ray of each type .
it is a nontrivial fact that a solution @xmath338 exists on the sector if the sector contains _ at most _ one stokes ray of each type ( this is the meaning of sufficiently narrow ) . a sector which contains exactly one stokes ray of each type
is called a stokes sector .
thus we have the fundamental principle ( theorem 1.4 of @xcite ) that there is a unique solution @xmath338 with asymptotic expansion @xmath320 on any stokes sector .
explicitly , the condition @xmath352 means that @xmath357 , i.e.@xmath358 , @xmath359 , and this simply means @xmath360 .
thus there are 8 stokes rays ; 2 of each of 4 types .
these are shown in fig .
[ stokesrays ]
. it will be convenient to regard @xmath362 as a subset of the universal covering surface @xmath363 ( rather than @xmath58 ) , so that we can define further subsets @xmath364 by @xmath365 the projection of this to @xmath58 is @xmath366 , which is also a stokes sector .
hence , for any @xmath367 , we have a unique holomorphic solution @xmath368 such that @xmath369 as @xmath370 in @xmath364 .
such a solution may be analytically continued to @xmath363 , and we shall use the same notation @xmath368 for the extended solution . on the overlap @xmath371 ( and hence on all of @xmath363 ) the two solutions @xmath372 must be related by a constant matrix .
we write @xmath373 and call @xmath374 the @xmath375-th stokes matrix ( at @xmath25 ) .
first , we note that the coefficients of @xmath320 are holomorphic on @xmath58 .
hence @xmath377 and @xmath378 have the same asymptotic expansion @xmath320 on @xmath364 .
it follows that they must be equal .
applying this fact twice , we obtain @xmath379 which gives @xmath380 , as required .
we shall in fact need a collection of stokes sectors compatible with the symmetries .
for this , we introduce the additional sectors @xmath381 i.e. we allow @xmath382 from now on , rather than @xmath367 .
let us write @xmath383 so that @xmath384 .
the cyclic symmetry @xmath321 of @xmath320 ( lemma [ formalsymmetries ] ) allows us to deduce that @xmath385 because the left and right hand sides have the same asymptotic expansion for @xmath388 ( and @xmath389 ) , hence must be equal
. similarly , the anti - symmetry and reality conditions follow from the corresponding properties of @xmath320 and the formulae @xmath390 , @xmath391 .
this is similar to the proof of uniqueness of @xmath368 .
namely , we have @xmath396 hence @xmath397 for all @xmath40 . moreover , for each @xmath394 such that @xmath346 , @xmath398 if there exists a path @xmath347 in @xmath399 such that @xmath350 .
since @xmath399 is a sector of angle @xmath356 , @xmath400 can be nonzero only if @xmath401 is equal to the right hand half plane @xmath402 .
this is equivalent to the condition @xmath403 mod @xmath356 .
we know that @xmath418 for some @xmath419 .
the reality condition ( at @xmath25 ) gives @xmath420 , @xmath421 .
the anti - symmetry condition gives @xmath422 in case 4a , and @xmath423 in case 4b .
the stated result follows .
[ charac ] the monodromy at @xmath25 is @xmath424 .
the characteristic polynomial of @xmath425 is @xmath426 if @xmath23 is a root of this polynomial , then so is @xmath427 in case 4a , and @xmath428 in case 4b .
the monodromy is ( the conjugacy class of ) @xmath429 . from the cyclic symmetry we have @xmath430 , and @xmath431 . since @xmath432
, we obtain @xmath433 .
the characteristic polynomial of @xmath425 may be computed from the formulae in the proposition .
the final statement follows from the anti - symmetry condition . concerning the connection matrix @xmath292
, we can argue indirectly that this is also determined by @xmath297 , because proposition [ injective ] ( below ) and theorem a say that @xmath297 determine @xmath298 and hence the functions @xmath440 .
( we compute @xmath292 in @xcite . ) * proof of theorem b(i ) : * for any @xmath441 in the region @xmath442 theorem a gives a solution @xmath440 of the system @xmath443 this corresponds to a solution @xmath444 of the system ( [ ost ] ) in cases 4a , 4b of table [ tableofcases ] .
our aim is to prove , in both cases , that @xmath445 we shall do this by computing the ( conjugacy class of the ) monodromy at @xmath25 in two ways .
first , by corollary [ charac ] , the monodromy is @xmath424 . on the other hand ,
the monodromy can be computed from the o.d.e .
@xmath446 by using the known asymptotic behaviour of @xmath440 .
namely , by letting @xmath447 , we obtain the constant coefficient o.d.e .
@xmath448 where @xmath449 in case 4a , and @xmath450 in case 4b .
we deduce that the eigenvalues of @xmath424 are @xmath451 , @xmath452 , @xmath453 , @xmath454 .
hence there exist integers @xmath455 such that the eigenvalues of @xmath425 are @xmath456 now , when @xmath457 for all @xmath40 , we must have @xmath458 for all @xmath40 , hence @xmath459 for all @xmath375 , and the eigenvalues of @xmath425 are @xmath308
. thus @xmath460 must be a permutation of @xmath461 . to compute @xmath455 we consider cases 4a , 4b separately . by the last statement of corollary [ charac ] ,
@xmath462 for some @xmath463 .
using proposition [ qk2 ] , and the fact that @xmath464 are the first and second elementary symmetric functions of the roots of the characteristic polynomial , we obtain @xmath465 and similarly @xmath466 of the eight values of @xmath467 such that @xmath468 , we shall show that only @xmath469 or @xmath470 are possible . to eliminate the other six cases @xmath471 we use the following fact : theorem
a implies that the map @xmath476 is bijective .
moreover , the proof shows that the dependence of @xmath477 on @xmath478 ( and that of @xmath479 on @xmath480 ) is monotone , hence the map @xmath481 is bijective as well .
on the other hand , proposition 2.2 of @xcite implies that the map @xmath482 is injective ; the failure of the riemann - hilbert map @xmath483 to be injective is entirely due to the isomonodromic deformation parameter @xmath34 .
we conclude that @xmath484 is injective , hence so is @xmath472 .
to examine injectivity of our eight candidates , it suffices to consider the expression @xmath485 evidently the map @xmath486 fails to be injective on the region @xmath487 in @xmath488\times [ 0,\pi]$ ] if @xmath489 fails to be injective on @xmath488 $ ] , or if @xmath490 fails to be injective on @xmath488 $ ] .
since @xmath491 is injective only on the intervals @xmath492 $ ] , this criterion eliminates @xmath493 and @xmath494 , hence all six cases listed above .
this time corollary [ charac ] shows that @xmath495 for some @xmath463 .
we obtain @xmath496 and @xmath497 this time @xmath498 and @xmath499 are eliminated as they would render the map @xmath500 non - injective .
we conclude that @xmath501 or @xmath502 .
again , we obtain the stated values for @xmath297 .
lemma [ formalsymmetries ] holds ( with @xmath508 instead of @xmath509 ) , but this time we have @xmath510 the matrix @xmath511 is @xmath512 the stokes rays are given by @xmath513 .
we choose @xmath514 as initial stokes sector , then @xmath515 , @xmath516 .
versions of lemmas [ psik ] and [ qk ] hold here with the following modifications : the analogue of lemma [ triangular ] is that , for @xmath346 , the @xmath394 entry of @xmath392 is @xmath24 if @xmath521 mod @xmath522 . thus the fundamental matrices @xmath523 have the form @xmath524 , @xmath525 where @xmath526 denotes the matrix whose only nonzero entry is @xmath393 in the @xmath394 position ( @xmath527 ) .
the reality condition gives @xmath528 , @xmath529 .
let us write @xmath530 then the anti - symmetry conditions imply that @xmath531 where @xmath415 are real .
finally , in analogy to corollary [ charac ] , the monodromy at @xmath25 is @xmath532 , and the characteristic polynomial of @xmath533 is @xmath534 if @xmath23 is a root of this polynomial , then so is @xmath427 in case 5a , and @xmath428 in case 5b .
* proof of theorem b(ii ) : * to compute @xmath535 in cases 5a , 5b we use the fact that the eigenvalues of @xmath532 are @xmath451 , @xmath452 , @xmath453 , @xmath454 , @xmath536 . hence there exist integers @xmath537 such that the eigenvalues of @xmath533 are @xmath538 to determine @xmath537 we consider each case separately . from the symmetry @xmath539
, we have @xmath540 for some @xmath541 . we obtain @xmath542 and similarly @xmath543 the analogue of proposition [ injective ]
is that the map @xmath472 is injective on the region @xmath473 , @xmath544 , @xmath475 . to investigate injectivity , it suffices to investigate @xmath545 now , the map @xmath546 is injective on @xmath547 $ ] only for @xmath548 .
the map @xmath549 is injective on @xmath550 $ ] only for @xmath551 .
we conclude that @xmath552 this gives the stated values for @xmath297 . the matrices @xmath392 and their cyclic / reality / anti - symmetry properties are the same as in cases 5a , 5b , except that in the anti - symmetry formula ( lemma [ qk ] ) we have @xmath559 this leads to @xmath560 with @xmath415 real .
if @xmath23 is a root of the characteristic polynomial of @xmath533 , then so is @xmath561 in case 5c , @xmath562 in case 5d , and @xmath563 in case 5e . in lemma [ formalsymmetries ]
we have @xmath581 the matrix @xmath511 is @xmath582 the stokes rays are given by @xmath583 .
we choose @xmath584 as initial stokes sector , then @xmath585 versions of lemmas [ psik ] and [ qk ] hold here with the following modifications : the analogue of lemma [ triangular ] is that , for @xmath346 , the @xmath394 entry of @xmath392 is @xmath24 if @xmath590 mod @xmath522 .
thus the fundamental matrices @xmath591 have the form @xmath592 , @xmath593 .
the reality conditions are @xmath594 , @xmath595 , @xmath596 .
let us write @xmath597 then the anti - symmetry conditions imply that @xmath598 and @xmath599 where @xmath415 are real .
the monodromy at @xmath25 is @xmath600 , and the characteristic polynomial of @xmath601 is @xmath602 if @xmath23 is a root of this polynomial , then so is @xmath561 in case 6a , @xmath562 in case 6b , and @xmath563 in case 6c .
* proof of theorem b(iv ) : * as the eigenvalues of @xmath600 are @xmath451 , @xmath452 , @xmath453 , @xmath454 , @xmath536 , @xmath603 , there exist integers @xmath604 such that the eigenvalues of @xmath601 are @xmath605 to determine @xmath604 we consider each case separately .
from the symmetry @xmath606 , we have @xmath607 for some @xmath608 .
we obtain @xmath609 and @xmath610 the analogue of proposition [ injective ] is that the map @xmath472 is injective on the region @xmath611 , @xmath544 , @xmath475 .
the map @xmath612 is injective on @xmath613 $ ] only for @xmath614 .
the map @xmath615 is injective on @xmath616 $ ] only for @xmath617 .
we conclude that @xmath469 or @xmath618 .
this gives the stated values for @xmath297 .
here @xmath619 for some @xmath608 .
this gives @xmath620 and @xmath621 the map @xmath622 is injective on @xmath613 $ ] only for @xmath623 .
the map @xmath624 is injective on @xmath616 $ ] only for @xmath625 .
hence @xmath501 or @xmath626 .
this leads to the same values for @xmath297 as in case 6a .
here @xmath627 for some @xmath608 .
we obtain @xmath628 and @xmath629 the map @xmath630 is injective on @xmath613 $ ] only for @xmath631 .
the map @xmath632 is injective on @xmath616 $ ] only for @xmath633 .
we conclude that @xmath634 or @xmath635 .
again this gives the same values for @xmath297 as in case 6a .
s. cecotti and c. vafa , _ exact results for supersymmetric @xmath636 models _ , phys .
* 68 * ( 1992 ) , 903906 .
s. cecotti and c. vafa , _ on classification of @xmath637 supersymmetric theories _ , commun .
* 158 * ( 1993 ) , 569644 .
v. cortes and l. schfer , _ topological - antitopological fusion equations , pluriharmonic maps and special khler manifolds _ , complex , contact and symmetric manifolds , progress in mathematics 234 , eds
. o. kowalski et al . ,
birkhuser , 2005 , pp .
j. dorfmeister , _ generalized weierstrass representations of surfaces _ , surveys on geometry and integrable systems , eds .
m. guest et al .
, advanced studies in pure math .
51 , math .
japan , 2008 , pp .
55111 .
j. f. dorfmeister , m. a. guest , and w. rossman , _ the tt * structure of the quantum cohomology of @xmath638 from the viewpoint of differential geometry _ , asian j. math . * 14 * ( 2010 ) 417438 .
b. dubrovin , _ geometry and integrability of topological - antitopological fusion _ ,
comm . math .
* 152 * ( 1993 ) , 539564 . m. a. guest and c .- s .
lin , _ some tt * structures and their integral stokes data _ , comm .
number theory phys . 6
( 2012 ) 785803 . c. hertling , _ tt * geometry , frobenius manifolds , their connections , and the construction for singularities _ , j. reine angew
* 555 * ( 2003 ) , 77161 .
m. jimbo , t. miwa , and k. ueno , _ monodromy preserving deformation of linear ordinary differential equations with rational coefficients i _ , physica d * 2 * ( 1981 ) 306352 ; ii : * 2 * ( 1981 ) 407448 ; iii : * 4 * ( 1981 ) 2646 .
l. katzarkov , m. kontsevich , and t. pantev , _ hodge theoretic aspects of mirror symmetry _ , from hodge theory to integrability and tqft : tt*-geometry , eds .
r. y. donagi and k. wendland , proc . of symp .
pure math .
78 , amer .
2007 , pp . 87174 .
i. mcintosh , _ special lagrangian cones in @xmath639 and primitive harmonic maps _ , j. lond .
* 67 * ( 2003 ) , 769789 . | we describe all smooth solutions of the two - function tt*-toda equations ( a version of the tt * equations , or equations for harmonic maps into @xmath0 ) in terms of ( i ) asymptotic data , ( ii ) holomorphic data , and ( iii ) monodromy data .
this allows us to find all solutions with integral stokes data .
these include solutions associated to nonlinear sigma models ( quantum cohomology ) or landau - ginzburg models ( unfoldings of singularities ) , as conjectured by cecotti and vafa .
in particular we establish the existence of a new family of pure and polarized terp structures in the sense of @xcite , or noncommutative variations of hodge structures in the sense of @xcite . |
transition - metal oxides with perovskite - based structure exhibit many fascinating phenomena such as colossal magnetoresistance in mn oxides @xcite , high temperature superconductivity in cu oxides @xcite , and ferroelectricity in ti oxides @xcite .
perovskite co oxides also show very rich physics such as spin - state transition @xcite , ferromagnetism @xcite , and thermoelectricity @xcite .
the ferromagnetism in the perovskite co oxide la@xmath0sr@xmath1coo@xmath6 has been explained in terms of the double - exchange interaction between co@xmath7 and co@xmath8 in the intermediate spin state .
the ferromagnetic transition temperature ( @xmath9 ) is 270 k at highest , which is significantly lower than @xmath9=370 k of perovskite mn oxides at an optimum composition . in recent years ,
the a - site ordered perovskite @xmath10baco@xmath11o@xmath12 ( @xmath13 rare earth element ) has been intensively studied @xcite .
it shows a ferromagnetic transition around 300 k ( higher than @xmath9 of la@xmath0sr@xmath1coo@xmath6 ) with giant magnetoresistance @xcite .
in addition , a transition from the ferromagnetic state to an antiferromagnetic state is seen around 255 k in the magnetization @xcite .
the complicated magnetism has been explained as follows : the co@xmath7 ions in the intermediate spin ( is ) state of @xmath14 exhibit an orbital ordering at 350 k along the @xmath15 axis @xcite , which induces ferromagnetic interaction along the @xmath15 axis . on the other hand ,
antiferromagnetic interaction is expected along the @xmath16 axis because of the conventional superexchange , which causes the antiferromagnetic order below 270 k. at higher temperatures , thermally excited carriers induce the double - exchange interaction to align the ferromagnetic chains , and stabilizes the ferromagnetic state from 270 to 300 k @xcite .
very recently , withers _
_ @xcite and istomin _ et al .
_ @xcite independently reported a new a - site ordered perovskite co oxide sr@xmath17coo@xmath2 ( @xmath10 = y and lanthanide ) . according to their structure analysis , sr and @xmath10
are ordered because of the different ionic radii .
the @xmath10 ions prefer to occupy at every two sites in the @xmath18 plane , which doubles the lattice parameter of the @xmath15 and @xmath16 axes from the primitive perovskite cell .
the ordered @xmath18 planes are stacked along the @xmath19-axis in units of four cells , and the @xmath19 axis length quadruples that of the primitive cell .
this will give an ideal composition of sr@xmath20coo@xmath2 , but a particular composition near @xmath31/3 has been extensively studied so far . aside from the cation ordering , the two groups proposed different oxygen ordering structures , and the complete structure is still controversial .
magnetic properties of sr@xmath0y@xmath21coo@xmath2 were measured only for @xmath22=0.1 and 0.33 @xcite , and we expected that the peculiar sr / y ordering near @xmath23 could show a novel magnetism . we have found that polycrystalline samples of sr@xmath0y@xmath1coo@xmath2 show a ferromagnetic transition at 335 k in a narrow range of @xmath24 , which is the highest transition temperature among perovskite co oxides .
the magnetization of the @xmath3 0.225 sample is 0.25 @xmath5/co at 10 k , indicating a bulk ferromagnet .
we suggest that this ferromagnetism is driven by the orbital ordering , just like the ferromagnetism in @xmath10baco@xmath11o@xmath12 .
however , @xmath9 of the former is even higher than that of the latter , and ferromagnetic order survives down to 5 k. we will discuss a possible origin for the difference .
polycrystalline samples of sr@xmath0y@xmath1coo@xmath2 ( @xmath3 0.1 , 0.15 , 0.2 , 0.215 , 0.225 , 0.25 , 0.3 and 0.4 ) and sr@xmath25coo@xmath2 ( @xmath13dy , ho , and er ) were prepared by a solid - state reaction .
stoichiometric amounts of srco@xmath6 , y@xmath11o@xmath6 , dy@xmath11o@xmath6 , ho@xmath11o@xmath6 , er@xmath11o@xmath6 , and co@xmath6o@xmath26 were mixed , and the mixture was calcined at 1100@xmath27c for 12h in air .
the product was finely ground , pressed into a pellet , and sintered at 1100@xmath27c for 48h in air , followed by slow cooling down to room temperature in the furnace .
that the resultant oxygen content of the sample represented an equilibrium value ( in air ) was confirmed by means of thermogravimetric annealing experiments .
the oxygen content was determined through iodometric titration@xcite .
a powdered sample of 20 - 40 mg was dissolved in 1 m hcl solution ( @xmath28 @xmath29 ml ) containing an excess of ki .
the titration of the formed i@xmath11 was performed against a standard 0.015 m sodium thiosulphate solution using starch as an indicator . for each sample
, the analysis was repeated three times in minimum with a reproducibility better than @xmath300.003 for @xmath31 . throughout this paper , we will use the nominal composition @xmath22 , although the third decimal point of @xmath22 includes some uncertainty .
we evaluated through energy dispersive x - ray analysis ( edx ) the real compositions for @xmath3 0.215 and 0.225 to be sr @xmath32
y @xmath33 and @xmath34 , respectively , meaning that @xmath22 is correct within an accuracy of @xmath350.01 . as a reference ,
sr@xmath36y@xmath37coo@xmath2 powder was sealed with kclo@xmath26 in a gold capsule , and then treated at 3 gpa and 700@xmath27c for 30 min in a conventional cubic anvil - type high pressure apparatus . the x - ray diffraction ( xrd ) of the sample
was measured using a standard diffractometer with fe k@xmath38 radiation as an x - ray source in the @xmath39-@xmath40 scan mode .
the resistivity was measured by a four - probe method , below room temperature ( 4.2 - 300 k ) in a liquid he cryostat , and above room temperature ( 300 - 400 k ) in an electronic furnace .
the magnetization @xmath41 was measured from 5 to 400 k by a commercial superconducting quantum interference device ( squid , quantum design mpms ) .
we applied 0.01 - 0.1 t for the measurement of @xmath41 ( @xmath42 ) data , and -7 to 7 t for @xmath41 ( @xmath43 ) data .
( @xmath42 ) ) of sr@xmath0y@xmath1coo@xmath2 ( @xmath3 0.1 , 0.15 , 0.2 , 0.215 , 0.225 , 0.25 , 0.3 and 0.4 ) , ( b ) magnetization versus magnetic field ( @xmath41(@xmath43 ) ) of sr@xmath36y@xmath37coo@xmath2 .
the insets of fig .
1(a ) and ( b ) show @xmath41(@xmath42 ) and @xmath41(@xmath43 ) of sr@xmath36y@xmath37coo@xmath6 , respectively.,width=264 ] figure 1(a ) shows the magnetization of sr@xmath0y@xmath1coo@xmath2 at 0.1 t as a function of temperature .
a tiny trace of a ferromagnetic transition is seen at 300 k for @xmath44 and 0.15 , and a weak and broad peak is seen near 300 k for @xmath45 and 0.4 , which is consistent with the literature @xcite .
surprisingly , the samples only in a narrow compositional range of @xmath24 show large magnetization .
the ferromagnetic transition temperature ( @xmath9 ) is 335 k , which is , to our knowledge , the highest temperature in perovskite co oxides .
the range of @xmath24 is quite singular , below which or above which the magnetization rapidly fades away .
this makes a remarkable contrast with la@xmath0sr@xmath1coo@xmath6 , where @xmath9 and magnetization continuously change with the sr content .
thus , the observed ferromagnetism should not be attributed to the double - exchange mechanism .
the inset of fig .
1(a ) shows @xmath41(@xmath42 ) of sr@xmath36y@xmath37coo@xmath6 annealed in high - pressure oxygen .
we expect no oxygen vacancies ( @xmath46 ) for this sample .
thus the formal valence of co is more than 3.7 , and the magnetism seems to be dominated by the double exchange interaction between co@xmath8 and co@xmath7 .
in fact , the observed ferromagnetic state is quite different from that in the oxygen - deficient samples : @xmath9 is remarkably lower , but the magnetization is larger .
these features are essentially similar to @xmath41(@xmath42 ) of la@xmath0sr@xmath1coo@xmath6 @xcite .
these differences clearly reveal that oxygen deficiency is essential to the peculiar room - temperature ferromagnetism .
figure 1(b ) shows the magnetic - field dependence of the magnetization ( @xmath41(@xmath43 ) ) of sr@xmath36y@xmath37coo@xmath2 at 10 k. at 7 t , the magnetization reaches 0.25 @xmath5/co .
although this value is somewhat smaller than that of la@xmath0sr@xmath1coo@xmath6 ( typically more than 1 @xmath5/co ) , it is large enough to be taken as an indication of bulk ferromagnetism , and is almost equal to that of polycrystalline sample of gdbaco@xmath11o@xmath12 at 250 k @xcite .
the inset of fig .
1(b ) shows @xmath41(@xmath43 ) of sr@xmath36y@xmath37coo@xmath6 ( the annealed sample ) at 10 k in which the @xmath41(@xmath43 ) curve is quantitatively different from that in the non - annealed samples .
these experimental facts further consolidate that the ferromagnetism in sr@xmath0y@xmath21coo@xmath2 is not caused by the double - exchange mechanism .
y@xmath37coo@xmath2 and ( b ) sr@xmath36er@xmath37coo@xmath2 .
( c ) x - ray diffraction patterns around 26@xmath27 of sr@xmath0y@xmath1coo@xmath2 ( @xmath3 0.1 , 0.15 , 0.2 , 0.225 , 0.3 and 0.4 ) and sr@xmath36er@xmath37coo@xmath2.,width=245 ] figure 2(a ) shows the xrd pattern of sr@xmath36y@xmath37coo@xmath2 , which is consistent with those of ref .
all the peaks are indexed as a tetragonal cell of the space group i4/mmm with the lattice parameters of @xmath47 and @xmath48 , except for small peaks of y@xmath11o@xmath6 ( at most 2% ; note that the intensity scale is logarithmic ) .
note that the ( @xmath4900 ) , ( 0@xmath500 ) , ( 00@xmath51 ) ( @xmath52 @xmath53 2@xmath54 ) peaks do not appear according to the extinction rule .
figure 2(b ) shows the xrd pattern of sr@xmath36er@xmath37coo@xmath2 , where all the peaks except for small peaks of er@xmath11o@xmath6 ( at most 4% ) are again successfully indexed as the tetragonal cell . the only difference from fig .
2(a ) is that the ( 101 ) , ( 110 ) and ( 103 ) peaks are much stronger in fig .
this should be associated with the different x - ray scattering cross sections of er and y , directly indicating that these superstructure peaks come from the a - site ordering . in other words , if the a site cations were disordered ( but the co or o ions were ordered ) , the two xrd patterns should be identical .
although the possibility of the ordering of the co and o ions is not excluded , the above results strongly support that the a - site ordering is predominant @xcite .
.crystallographic and iodometric titration data for sr@xmath0y@xmath1coo@xmath2 ( @xmath3 0.1 , 0.15 , 0.2 , 0.225 , 0.25 , 0.3 and 0.4 ) .
standard deviation is given in parentheses . [
cols="<,^,^,^,^,^",options="header " , ] [ ionradi ] figure 2(c ) shows the @xmath22 dependence of the ( 103 ) peak , which is a direct probe for the a - site ordering .
clearly , it appears above @xmath55 , which exactly corresponds to the emergence of the room - temperature ferromagnetism in this compound .
as listed in table i , the crystal symmetry for @xmath44 and 0.15 is cubic ( pm3 m ) without any superstructure peaks like the ( 103 ) peak in fig .
2(a ) , which is consistent with ref .
thus we conclude that the a - site ordering drives the ferromagnetism in the co ions .
this is the same situation as in the case of @xmath10baco@xmath11o@xmath12 , where the ordering of @xmath10 and ba stabilizes the co@xmath7 ordering .
thus we think that the ferromagnetism in the present material would be driven by the orbital ordering , although the ordering pattern may be different because @xmath10baco@xmath11o@xmath12 shows a different xrd pattern .
we further note that the peak is monotonically shifted to a higher degree with increasing @xmath22 , indicating that the lattice parameter is systematically decreased by substitution of y for sr ( see table i ) . on first glance , it looks inconsistent that the @xmath45 sample does not show the bulk ferromagnetism in spite of having the same intensity of the ( 103 ) peak as the @xmath56 sample .
we think that the ferromagnetism is quite susceptible to the sr / y solid solution in the sr site : y can not help occupying the sr site for @xmath571/4 .
in fact , our conjecture is verified by the fact that sr@xmath58ca@xmath59y@xmath60coo@xmath2 rapidly loses ferromagnetism with @xmath61 @xcite .
in contrast , the oxygen content is unlikely to play an important role , because it remains essentially constant ( @xmath62 ) between @xmath56 and 0.3 .
coo@xmath2 ( @xmath10 = dy , ho , and er ) .
the inset shows 1/@xmath41 of sr@xmath36y@xmath37coo@xmath2.,width=245 ] figure 3 shows the @xmath41(@xmath42 ) of sr@xmath25coo@xmath2 ( @xmath10 = dy , ho and er ) .
not only the y - substituted samples but also the other compounds exhibit a similar ferromagnetism below 335 k. this means that the magnetic moments of the lanthanides are irrelevant to the 335 k transition .
the inset of fig .
3 shows 1/@xmath41 of sr@xmath36y@xmath37coo@xmath2 , in which a kink is observed at 360 k. a similar kink was observed in 1/@xmath41 of tbbaco@xmath11o@xmath12 at the orbital - ordering temperature @xcite .
the low - temperature magnetization is affected by the magnetic moment of dy@xmath7 , ho@xmath7 , and er@xmath7 .
the low - temperature susceptibility of the er- and dy - substituted samples obeys the curie law below about 50 k. the effective magnetic moment is evaluated to be 6.7 @xmath63/er for the er - substituted sample , corresponding to 70 % of that expected from er@xmath7 of @xmath64 and @xmath65 , where @xmath66 is the total angular momentum and @xmath67 is lande s g factor .
the effective magnetic moment is 5.3 @xmath63/dy for the dy - substituted sample , corresponding to 50 % of that expected from dy@xmath7 ( @xmath6815/2 and @xmath694/3 ) .
since the orbital angular momentum is sometimes quenched in the presence of ligand field , the magnetic moments are reasonably evaluated for the above samples .
on the other hand , the ho - substituted sample shows anomalous susceptibility that decreases with decreasing temperature below 50 k. it may indicate an antiferromagnetic coupling between co@xmath7 and ho@xmath7 ions , and such a 3@xmath70 - 4@xmath71 coupling is actually observed in @xmath72cu@xmath11o@xmath73 @xcite . figure 4 shows the resistivity of sr@xmath0y@xmath1coo@xmath2 . the temperature dependence of resistivity is nonmetallic , which is again incompatible with the double - exchange mechanism . for @xmath24 samples , a resistivity jump is seen at 360 k , which is close to @xmath9 .
a similar but larger jump was reported in the resistivity of gdbaco@xmath11o@xmath12 @xcite .
we further note that the resistivity for 0.2 @xmath74 0.25 is clearly distinguished from that for @xmath75 0.15 or @xmath76 0.3 .
the inset of fig .
4 shows the resistivity of sr@xmath36y@xmath37coo@xmath6 annealed in the high - pressure oxygen .
the resistivity shows metallic conduction with a kink at 250 k , which is very different from the resistivity of the non - annealed samples .
now we discuss the origin of the room - temperature ferromagnetism in sr@xmath0y@xmath1coo@xmath2 ( @xmath24 ) .
as listed in table i , the oxygen content 3-@xmath31 for @xmath3 0.25 is @xmath77 2.64 , corresponding to the formal co valence of @xmath773.03 .
this is in between the oxygen content of 2.69 for @xmath30.2 by james et al . @xcite and the oxygen content of 2.62 for @xmath30.3 by istomin et al .
@xcite . very recently ,
maignan _ et al .
_ @xcite found a similar oxygen content of 2.66 for @xmath22=0.33 , and reported that the formal valence of co is 2.99 . from these data
we believe that the co ions are essentially trivalent for @xmath3 0.25 .
thus the ferromagnetism most likely arises from the orbital ordering of the is co@xmath7 ions , as in the case of @xmath10baco@xmath11o@xmath12 , which naturally explains the several similarities to the ferromagnetism in @xmath10baco@xmath11o@xmath12 mentioned above .
y@xmath1coo@xmath2 ( @xmath3 0.1 , 0.15 , 0.2 , 0.225 , 0.25 , 0.3 , and 0.4 ) .
the arrow indicates the kink temperature in the inset of fig .
3 . the inset shows the resistivity of sr@xmath36y@xmath37coo@xmath6 the arrow indicates the curie temperature.,width=245 ] next , we will compare sr@xmath0y@xmath1coo@xmath2 with @xmath10baco@xmath11o@xmath12 .
the ferromagnetic order is highly robust down to 5 k in sr@xmath0y@xmath1coo@xmath2 , and is most likely the ground state of this system .
according to the theory of orbital ordering @xcite , an antiferro - type orbital order favors a ferromagnetic order , and a ferro - type orbital order favors an antiferromagnetic order .
suppose the present compound be an insulator consisting of is co@xmath7 .
then an orbital order will be antiferro - type , and prevent formation of the @xmath78 band , which drives a ferromagnetic order .
similar ferromagnetic states are seen in other orbital - ordered systems , such as k@xmath11cuf@xmath26 @xcite and ytio@xmath6 @xcite . in this context
@xmath10baco@xmath11o@xmath12 is rather exceptional , in which the antiferromagnetic order is the ground state in spite of the anitiferro - type orbital ordering .
thus the ferromagnetic state competes with the antiferromagnetic state in @xmath10baco@xmath11o@xmath12 , which causes large magnetoresistance .
such a competition can explain why @xmath9 is lower than @xmath9 of sr@xmath0y@xmath1coo@xmath2 .
finally , we would like to add a few notes .
( i ) as a preliminary measurement , we measured the magnetoresistance of sr@xmath36y@xmath37coo@xmath79 at 300 k. the value is @xmath80 % at 7 t , which is very small compared with that of a typical giant magnetoresistance material .
this suggests that the ground state is ferromagnetic and insulating , which is consistent with our orbital - ordering scenario .
( ii ) we examined an impurity effect , and observed that the ferromagnetism is quite susceptible against impurities , where only a 1%-substitution of mn for co destroys a half of the saturation magnetization .
such result can not be understood from simple dilution effects , and could be attributed to the impurity effect on the orbital ordering . in a case of charge ordering , a tiny disorder strongly suppresses the charge ordering state @xcite .
we have measured magnetic susceptibility and resistivity of sr@xmath0y@xmath1coo@xmath2 ( @xmath3 0 , 0.1 , 0.15 , 0.2 , 0.225 , 0.25 , 0.3 and 0.4 ) , and have found that sr@xmath0y@xmath1coo@xmath2 for 0.2 @xmath4 0.25 can be regarded as a room - temperature ferromagnet .
the ferromagnetic transition temperature is approximately 335 k , which is the highest transition temperature among perovskite co oxides .
the magnetization of the @xmath3 0.225 sample at 10 k is 0.25 @xmath5/co , which implies a bulk ferromagnet .
we conclude that the ferromagnetic order is driven by the peculiar y / sr ordering .
we expect that this material will open a new window to materials science of the a - site ordered co oxides .
marc respaud , carlos frontera , jos luis garca - munoz , miguel ngel g. aranda , bertrand raquet , jean marc broto , harison rakoto , michel goiran , anna llobet , and juan rodrguez - carvajal , phys .
b * 64 * , 214401 ( 2001 ) .
preliminary , we annealed the samples of @xmath30.25 and 0.3 in an o@xmath11 flow at 800@xmath27c , and found the magnetism essentially unchanged after the annealing . according to istomin et al .
( fig . 13 in ref .
@xcite ) , the 800@xmath27c annealing could decrease the oxygen content to some extent ( 1% ) , which suggests that the observed ferromagnetism is less dependent on oxygen content . | we have measured magnetic susceptibility and resistivity of sr@xmath0y@xmath1coo@xmath2 ( @xmath3 0.1 , 0.15 , 0.2 , 0.215 , 0.225 , 0.25 , 0.3 , and 0.4 ) , and found that sr@xmath0y@xmath1coo@xmath2 is a room temperature ferromagnet with a curie temperature of 335 k in a narrow compositional range of 0.2 @xmath4 0.25 .
this is the highest transition temperature among perovskite co oxides .
the saturation magnetization for @xmath3 0.225 is 0.25 @xmath5/co at 10 k , which implies that the observed ferromagnetism is a bulk effect .
we attribute this ferromagnetism to a peculiar sr / y ordering . |
consider the setup in fig . [
figure : setup ] .
the circular channel around the hole is an edge of a 2d ti and therefore hosts helical edge states .
we assume for simplicity that the spin - polarization axis is perpendicular to the plane of the ti .
the hamiltonian describing the two counter - propagating edge channels is then given by @xmath6 where arrows indicate the spin direction .
in the ground state no net current is carried from one side to the other , since the system is symmetric under a rotation of @xmath7 around the the @xmath8-axis orthogonal to the ti . however ,
if a voltage is applied across the circuit , electrons can start to flow from one side to the other , say from the left to the right .
this current will be proportional to the transfer matrix of the states that are occupied at the left side , but unoccupied on the right .
we therefore begin by calculating this transfer matrix .
when considering processes that transfers electrons from the left to the right , we can , because of the helicity of the edge states , restrict ourselves to up - spins along the upper edge , and down - spins along the lower edge .
further , we introduce the coordinate @xmath9 and @xmath10 along the upper and lower edges , respectively .
the eigenvalue equations along the two edges are then @xmath11 and the corresponding eigenstates can be written as @xmath12e^{-ipx_1/\hbar},\nonumber\\ |\psi_{\downarrow p}\rangle = & \left[\begin{array}{c } 0\\ 1 \end{array}\right]e^{-ipx_2/\hbar}.\nonumber\end{aligned}\ ] ] we now thread a magnetic flux of magnetic field strength @xmath13 through the hole . to describe this we choose the vector potential @xmath14 , which translates into @xmath15 and @xmath16 in the new @xmath17-coordinates .
the addition of this vector potential acts on the phase of the eigenstates according to @xmath12e^{-i(p + qbr/2)x_1/\hbar},\nonumber\\ |\psi_{\downarrow p}\rangle = & \left[\begin{array}{c } 0\\ 1 \end{array}\right]e^{-i(p - qbr/2)x_2/\hbar},\nonumber\end{aligned}\ ] ] where @xmath4 is the unit of electric charge .
it is therefore clear that the transfer matrix , which describes the transport of spins from the left side , @xmath18 , to the right side , @xmath19 , is given by @xmath20.\end{aligned}\ ] ] we here note that under a gauge transformation @xmath21 , where @xmath22 satisfies @xmath23 , the transfer matrix transforms as @xmath24 .
we have confirmed that this additional phase drops out of all physical quantities below , proving the gauge invariance of our results , and we can therefore set @xmath25 .
similarly , the overall phase in the above equation will drop out of all physical quantities .
this also justifies us in not having specified the chemical potential . because , as long as the spectrum is described by the same edge hamiltonian , the only role of the chemical potential is to determine around what momentum @xmath26 the relevant excitations are located .
the total transfer matrix for the system will not only depend on the transfer matrix that describes the motion around the hole , but also on the matrices that describe the transfer processes between the leads and the circular edge .
we will here assume that this process preserves phase coherence between the states in the leads and the ti edge states , and that it is described by a single tunneling parameter @xmath27 , which we for now set to @xmath28 to indicate perfect transmission between lead and edge .
that is , the transmission is described by the identity matrix , and therefore contributes trivially to the total transfer matrix .
however , we will in what follows be interested in tilting the ti by an angle @xmath29 relative to the quantization axis of the leads .
it is therefore necessary to also let the total transfer matrix encode a change of basis between the leads and the ti . for this purpose
we define two sets of coordinate axes , the laboratory axes @xmath30 , and the ti axes @xmath31 .
we choose to describe the electrons in the leads with the coordinates in the laboratory frame , while the edge states in the ti are described by the primed coordinates .
it is clear that eq
. refers to the transfer of states in the primed basis . in particular , we choose the @xmath32-axes along the direction of motion of the electrons through the circuit , while the @xmath33-axes are chosen such that they coincide when @xmath34 and @xmath35 is always perpendicular to the ti .
explicitly , the @xmath30- and @xmath31-coordinates are related through @xmath36 the change of basis from the @xmath30-basis to the @xmath31-basis for the spins is then given by @xmath37.\end{aligned}\ ] ] we have here used @xmath38 and @xmath39 to denote the transformations from the unprimed to the primed coordinates , and the primed to the unprimed coordinates , respectively .
the symbols @xmath38 and @xmath39 are chosen since they are applied at the left and right end of the system , respectively . with these definitions we are now ready to write down the complete transfer matrix for the system
@xmath40 here we have made explicit the dependence of @xmath41 on the parameters @xmath13 and @xmath2 from eq . , and on @xmath29 from eq . .
the main advantage of introducing the @xmath38 and @xmath39 matrices is that they allow us to work in the laboratory frame alone . to calculate the probability that an incoming spin @xmath42 in the left lead is transferred to a spin @xmath43 in the right lead , we now simply need to calculate the square of the corresponding matrix element @xmath44
as a first example of a concrete application , we consider a system with fully spin - polarized ferromagnetic leads only containing electrons with spin - up .
further , the sg spin - splitter is assumed to be oriented at an angle @xmath45 , which forces the incoming spins to split equally into both channels . because the leads only conduct spin - up electrons ,
the only relevant matrix element for the scattering matrix is @xmath46 the conductance is therefore given by @xmath47 it is clear that the prominent dependence of the current on the magnetic flux @xmath48 makes this setup ideal for measuring magnetic field strength , as a potential alternative to superconducting quantum interference devices ( squids ) .
the measurement resolution is directly set by the radius of the hole in the ti .
this is of special interest because it provides a potential route for high - resolution magnetic field measurements even at room temperatures.@xcite
next we note that the configuration in the previous section can also be used as a spintronics switch , with voltage used to encode 0 and 1 .
the two leads can be used as source and drain , while the magnetic field is used as the gate . from eq .
it is clear that a magnetic field strength @xmath49 corresponds to on and off states for @xmath50 even and odd , respectively , and we therefore define the magnetic switching quantum @xmath51 an alternative way to encode 1 and 0 is to use the currents of up- and down - spins , respectively .
this requires normal leads through which both up- and down - spins can be transported .
we therefore consider the same configuration , but now evaluate all four components of the transfer matrix @xmath52 : @xmath53 similarly as above , the square of the transfer matrices gives the transfer probability of the spin - polarized currents .
in particular , the off - diagonal matrix elements @xmath54 converts between up and down spin currents . therefore , the device relates the in and outgoing spin currents to each other through @xmath55 once again considering the special cases @xmath49 , with @xmath50 an integer , the currents transforms according to @xmath56 this means that the device can be switched between a normal lead and a not - gate , simply by changing @xmath13 by the switching quantum in eq . .
having seen how an sg ti device can be used to construct classical logic gates for spintronics , we finally turn to possible applications in quantum computing .
it has been shown that a universal quantum computer can be built using only two - qubit cnot - gates and single - qubit @xmath0-gates @xcite .
we here show that a sg ti spin - splitter provides a route for implementing the later of these two gates . for this purpose
we consider three sequential spin - splitters connected by normal leads .
the three devices are oriented as in fig .
[ figure : setup_qc ] , with the middle device oriented at an angle @xmath57 , while the first and last spin - splitters are at an angle @xmath58 . the total transfer matrix for the complete system is then given by @xmath59 when evaluated , this expression can be written as @xmath60\left[\begin{array}{cc } e^{i\beta_{3}/2 } & 0\\ 0 & e^{-i\beta_{3}/2 } \end{array}\right]\left[\begin{array}{cc } \cos(\frac{\beta_{2}}{2 } ) & \sin(\frac{\beta_{2}}{2})\\ -\sin(\frac{\beta_{2}}{2 } ) & \cos(\frac{\beta_{2}}{2 } ) \end{array}\right]\left[\begin{array}{cc } e^{i\beta_{1}/2 } & 0\\ 0 & e^{-i\beta_{1}/2 } \end{array}\right ] , \end{aligned}\ ] ] where @xmath61 . the six physical parameters @xmath62 are more than sufficient to make the four parameters @xmath63 and @xmath64 independent of each other .
moreover , when all these four parameters can be chosen independently , it is possible to express any @xmath65-matrix using eq .
( [ equation : qc_transfer_function ] ) @xcite .
thus , it is possible to implement any unitary single - qubit gate , and in particular any @xmath0-gate , through the use of three sequential solid state sg spin - splitters .
in fact , the overall @xmath5-phase provided by the parameter @xmath66 can be ignored for similar reasons that the @xmath5-phase provided by the gauge transformation @xmath21 can be ignored .
this phase would only be relevant if the incoming electron is further split up into one part passing through the device , and one part moving through another path joining only at the far right outgoing lead . in light of this
it is useful to think of the devices as exhibiting a @xmath0-ab effect . while the ordinary ab effect arise as a consequence of interference in a single @xmath5-phase , these devices relies on a generalized @xmath0-interference effect in the relative phase and amplitude of the up and down components of the spin . to be able to create an arbitrary @xmath0-transformation ,
a sequence of three devices is needed , while an individual spin - splitter gives rise to a subset of such @xmath0-transformations .
finally , we note that in this calculation we have omitted transfer matrices describing the propagation through the leads .
we are justified in doing so because these would be proportional to the identity and therefore only contribute to the irrelevant @xmath66 phase .
we would like to end with a few comments on some of the assumptions made .
first of all the tunnelling parameter @xmath27 which otherwise would have multiplied the @xmath38 and @xmath39 matrices was set to @xmath28 .
it is clear that the zeroth order correction to deviations from @xmath28 is to include the factor @xmath67 in front of all transmission coefficients , which shows up as @xmath68 in the conductivity .
the higher order corrections would come from particles which are reflected and travels an additional time around the loop . while such terms can introduce corrections to the interference pattern for intermediate field strengths
, they would not affect the result at multiples of the switching quantum in eq . .
the reason being that additional circuits around the loop will affect the relative phase between the up and down spins by multiples of @xmath69 .
such interference effect could also play a role for @xmath28 when ferromagnetic leads are used , because the down spins at the right edge will be completely reflected .
this can be solved by adding another ferromagnetic lead for the down spins to escape through at the right edge .
we also mention that although the setup in fig .
[ figure : setup_qc ] might seem prohibitively difficult to realize in practice , the focus has here been on providing a conceptually simple explanation of the phenomenon . in fact , the only reason the middle spin - splitter is tilted at an angle @xmath70 is to make its edge states have their spin - polarization perpendicular to those of the other two . in practice
it would therefore be possible to have all three devices in the same plane , if it is constructed out of two different types of 2d tis with perpendicular spin - polarization axes .
we have shown that the helical edge states of a 2d ti can be utilized to construct a solid state sg spin - splitter , which when threaded by a magnetic flux gives rise to a generalized @xmath0-ab interference effect .
with two ferromagnetic leads , the device can be used to accurately measure magnetic flux , as well as be used as a magnetic field gated spintronics switch . instead using normal leads , a switchable spintronics not - gate
can be implemented , or when using three devices connected in sequence , a @xmath0-gate for quantum computing is achieved .
we are grateful to j. bardarson , j. fransson , and e. sjqvist for useful discussions and the swedish research council ( vetenskapsrdet ) , the gran gustafsson foundation , and the swedish foundation for strategic research ( ssf ) for financial support . 9 see , e.g. , j. j. sakurai , _ modern quantum mechanics ( revised edition ) _ , addison - wesley publishing company , inc .
y. aharonov and d. bohm , phys . rev . *
115 * , 485 ( 1959 ) . c. l. kane and e. j. mele , phys .
lett . * 95 * , 226801 ( 2005 ) . c. l. kane and e. j. mele , phys .
95 * , 146802 ( 2005 ) .
b. a. bernevig and s .- c .
zhang , phys .
lett . * 96 * , 106802 ( 2006 ) . c. l. kane and e. j. mele , science * 314 * , 1692 ( 2006 ) .
b. a. bernevig , t. l. hughes , and s .- c .
zhang , science * 314 * , 1757 ( 2006 ) .
m. knig , s. wiedmann , c. brne , a. roth , h. buhmann , l. molenkamp , x. l. qi , and s .- c .
zhang , science * 318 * , 766 ( 2007 ) .
liu , w. feng , and y. yao , phys .
. lett . * 107 * , 076802 ( 2011 ) .
m. z. hasan and c. l. kane , rev .
phys . * 82 * , 3045 ( 2010 ) .
x. l. qi and s .- c .
zhang , rev .
* 83 * , 1057 ( 2011 ) .
y. zhang and a. vishwanath , phys .
lett . * 105 * , 206601 ( 2010 ) .
j. h. bardarson , p. w. brouwer , and j. e. moore , phys .
* 105 * , 156803 ( 2010 ) . h. peng , k. lai , d. kong , s. meister , y. chen , x .-
qi , s .- c .
zhang , z .- x .
shen , and y. cui , nature mat * 9 * , 225 ( 2010 ) .
s. cho , b. dellabetta , r. zhong , j. schneeloch , t. liu , g. gu , m. j. gilbert , and n. mason , nature comms .
* 6 * , 7634 ( 2015 ) .
r. ilan , f. de juan , and j. e. moore , arxiv:1410.5823 ( 2014 ) .
a. barenco , et al .
a. * 52 * , 52 ( 1995 ) .
y. xu , b. yan , h .- j .
zhang , j. wang , g. xu , p. tang , w. duan , and s .- c .
zhang , phys .
. lett . * 111 * , 136804 ( 2013 ) .
zhu , w .- j .
chen , y. xu , c .-
guan , c .- h .
liu , d. qian , s .- c .
zhang , and j .- f .
jia , nature mat .
http://dx.doi.org/10.1038/nmat4384 ( 2015 ) . | we show that the edge of a two - dimensional topological insulator can be used to construct a solid state stern - gerlach spin - splitter . by threading such a stern - gerlach apparatus with a magnetic flux ,
ahranov - bohm like interference effects are introduced . using ferromagnetic leads ,
the setup can be used to both measure magnetic flux and as a spintronics switch . with normal
metallic leads a switchable spintronics not - gate can be implemented .
furthermore , we show that a sequence of such devices can be used to construct a single - qubit @xmath0-gate , one of the two gates required for a universal quantum computer .
the field sensitivity , or switching field , @xmath1 is related to the device characteristic size @xmath2 through @xmath3 , with @xmath4 the unit of electric charge . two famous examples of the fundamental difference between quantum mechanical and classical particles are provided through the stern - gerlach ( sg ) experiment @xcite and the aharanov - bohm ( ab ) effect @xcite . the sg experiment demonstrates the peculiar behavior of the quantum mechanical spin , teaching us that for any chosen axis the spin can be pointing either up or down .
even more unintuitive , the spin can also be in a superposition of these two states , and thereby split in a sg apparatus to travel along different paths @xcite .
the ab effect , on the other hand , shows that the introduction of a magnetic vector potential has important effects on the phase of the wave function .
this is not merely a mathematical formality , but has measurable consequences in interference measurements .
when a particle travels along two different paths that enclose a magnetic flux , it picks up different phases along the two paths , even though the paths do not pass through the magnetic flux @xcite .
a topological insulator is a material with insulating bulk , but with topologically protected helical edge states .
here we show that it is possible to construct a solid state sg apparatus , or spin - splitter , using the edge states in a two - dimensional topological insulator ( 2d ti ) @xcite .
the device consists of a small hole drilled in the 2d ti , contacted by two leads . by threading a magnetic flux through the hole ,
an ab - like effect gives rise to important interference effects , which allows for precise manipulation of spin currents .
while the ordinary ab effect arises because of interference in a single complex number , the effects achieved here relies on modifying the relative phase between the up and down components of the spin .
thus , the effects described here can be classified as @xmath0-ab effects , while the ordinary situation corresponds to an @xmath5-ab effect .
while , the ab effect has recently attracted some attention in 3d ti @xcite , we here outline several concrete and different applications of the @xmath0-ab effect in 2d ti .
more specifically , we find that if using ferromagnetic leads , the device can be used for sensitive measurements of magnetic field strengths .
the same setup can also be used to implement a spintronic switch . instead using normal metallic leads , a switchable not spintronics gate
can easily be constructed .
finally , we also demonstrate how a sequential setup of normal lead solid state sg spin - splitters can be used to construct a single - qubit @xmath0-gate , one of two gates required to construct a universal quantum computer @xcite . |
experiments at the relativistic heavy - ion collider ( rhic ) suggest that the matter created in semi-/central au - au collisions at @xmath0@xmath1@xmath2 constitutes an equilibrated , strongly interacting medium at high energy density , well above the critical one at the expected phase boundary , @xmath3@xmath4 1gev/@xmath5 ( cf .
the experimental assessment papers@xcite and references therein ) .
this conclusion is essentially based on three evidences : ( i ) at low transverse momenta , @xmath6@xmath72 - 3gev ( comprising @xmath899% of the produced hadrons ) , ideal relativistic hydrodynamics describes well the single - particle spectra and their azimuthal asymmetry , @xmath9 ; the underlying collective expansion follows from kinetic equilibration times of around @[email protected]/@xmath11 , implying rapid ( local ) thermalization of the system at initial energy densities of about @xmath12@xmath430gev/@xmath13 .
( ii ) at high momenta , @xmath6@xmath145 - 6gev , the production of hadrons is suppressed by a factor of 4 - 5 relative to elementary @xmath15-@xmath15 ( or d - au ) collisions .
this has been interpreted as energy loss of energetic partons traveling through an almost opaque high gluon - density medium .
( iii ) at intermediate momenta , @xmath6@xmath42 - 6gev , surprisingly large baryon - to - meson ratios , as well as an empirical scaling of the hadron @xmath16 according to its constituent quark content , has been interpreted as quark coalescence as the prevalent hadronization mechanism in this regime , and thus as partonic degrees of freedom being active .
the microscopic mechanisms underlying thermalization and energy loss , however , are not well understood yet .
in fact , the jet quenching of light partons is so strong that high-@xmath6 particle emission appears to be mostly limited to the surface of the fireball , rendering a precise determination of the corresponding transport coefficient ( and maximal energy density ) difficult . while earlier approaches assumed the dominance of radiative energy loss via medium - induced gluon radiation ( which prevails in the limit of high jet energies ) cf .
@xcite , it has been realized subsequently that elastic ( 2@xmath172 ) scattering processes can not be neglected at currently accessible @xmath6@xmath710 - 15gev at rhic .
moreover , none of the evidences ( i)-(iii ) is directly connected to the fundamental properties distinguishing the quark - gluon plasma ( qgp ) from hadronic matter , namely the deconfinement of color charges and the restoration of the spontaneously broken chiral symmetry ( which is believed to generate most of the visible mass in the universe ) .
heavy - quark ( hq ) observables are expected to provide new insights into the aforementioned problems ( with the exception of chiral symmetry restoration ) . on the one hand ,
charm ( @xmath11 ) and bottom ( @xmath18 ) quarks , due to their relatively large mass , should suffer less energy loss and undergo delayed thermalization when traversing the qgp , and thus be more sensitive to the interactions with thermal partons than light quarks and gluons .
therefore , it came as a surprise when the measurement of single - electron ( @xmath19 ) spectra associated with the semileptonic decay of open - charm ( and -bottom ) hadrons showed a factor 4 - 5 suppression@xcite , very comparable to what has been found for pions .
in addition , the observed values for the @xmath19 elliptic flow , @xmath20 , reach up to 10% at @xmath6@xmath42gev , indicating the build - up of substantial _ early _ collectivity of @xmath11 quarks .
these observations reinforced the implementation of elastic energy - loss processes into the theoretical description of the spectra@xcite .
on the other hand , hq bound states ( quarkonia ) , due to their small size / large binding , are suitable probes of a surrounding medium of sufficiently high density .
e.g. , for a typical ( ground - state ) charmonium of size @[email protected] , the relevant parton density at which significant modifications are to be expected is @xmath22@xmath23@xmath24@xmath23@xmath25-@xmath26@xmath27 , which for an ideal qgp ( with @[email protected] massless flavors ) translates into a temperature of @xmath29@xmath23270mev@xmath23@xmath30 .
indeed , recent lattice qcd ( lqcd ) calculations indicate that @xmath31 and @xmath32 states survive in a qgp up to @xmath82@xmath33 .
the theoretical challenge is then to disentangle and quantify ( a ) medium modifications of the binding potential ( e.g. , color screening ) , ( b ) parton - induced dissociation reactions , and ( c ) in - medium changes of the hq mass which affects both the bound - state mass and its decay threshold , as well as to identify and establish connections to observables in heavy - ion collisions . in the first part of the talk ( sec .
[ sec_open ] ) , we evaluate hq diffusion and thermalization in the qgp employing both perturbative and _ nonperturbative _ elastic interactions . this problem is particularly suited to a fokker - planck approach for brownian motion in a thermal background@xcite , which allows to describe both the quasi - thermal and kinetic regime , and its transition .
a langevin simulation for @xmath11 and @xmath18 quarks in an expanding qgp at rhic is supplemented with coalescence ( and fragmentation ) at @xmath33 , thus implementing all of the three main features ( i)-(iii ) of the initial rhic data . in the second part of the talk ( sec .
[ sec_hidden ] ) , we discuss in - medium properties of heavy quarkonia , including information from lqcd .
we outline a @xmath29-matrix approach in which hq potentials from lqcd are used as input to simultaneously describe bound and scattering states .
this enables a comprehensive evaluation of euclidean correlation functions , which in turn can be checked against rather accurate results from lqcd .
we briefly discuss how in - medium quarkonium properties reflect themselves in observables in ultrarelativistic heavy - ion collisions .
[ sec_concl ] contains our conclusions .
recent calculations of hadronic spectral functions in a qgp , both within lqcd and lqcd - based potential models , suggest that resonance ( or bound ) states persist up to @xmath34 , for both heavy- ( @xmath35 ) and light - quark ( @xmath36 ) systems .
we conjectured@xcite that this also holds for heavy - light systems , and therefore could lead to significantly faster thermalization for @xmath11 and @xmath18 quarks as compared to perturbative qcd ( pqcd ) processes .
even at @xmath29=350mev ( which roughly corresponds to initial temperatures at rhic ) and for @[email protected] , elastic pqcd scattering , which is dominated by @xmath38-channel gluon exchange , results in a thermal relaxation time for @xmath11 quarks above 10fm/@xmath11 , well above a the typical qgp lifetime of @xmath39@xmath85fm/@xmath11 .
recent data on single - electron spectra associated with semileptonic heavy - meson decays have corroborated the need for nonperturbative hq interactions in the qgp .
as shown in fig .
[ fig_pqcd ] , perturbative energy - loss calculations appear insufficient to describe the strong suppression in the @xmath6-spectra ( left panel , using both elastic and radiative interactions ) and the large @xmath9 ( right panel , using radiative energy loss with a substantially upscaled transport coefficient ) .
while the plots indicate a ( lower ) applicability limit of the pqcd energy - loss approach of @xmath40@xmath42gev , we will argue below that nonperturbative effects may be relevant ( or even dominant ) up to at least @xmath40@xmath45gev .
in particular , radiative energy - loss calculations typically do not include the backward reactions ( detailed balance ) which are essential for building up collective behavior of the heavy quarks in the expanding thermal medium . as mentioned above , in this context
a brownian - motion approach is well suited to address the thermalization of the heavy quarks@xcite .
upon expanding the boltzmann equation in small momentum transfers , one arrives at a fokker - planck equation for the hq distribution function , @xmath41 , @xmath42 with ( momentum ) drag ( @xmath43 ) and diffusion ( @xmath44 ) constants ( related via @xmath45 ) .
the latter are calculated from corresponding matrix elements for hq scattering off light partons .
+ as mentioned above , our main ingredient here are resonance - mediated elastic interactions in @xmath46- and @xmath47-channel ( cf .
left panel of fig .
[ fig_tau ] ) which we model with an effective lagrangian according to@xcite @xmath48 @xmath49 is the resonance field representing `` @xmath50 '' and `` @xmath51 '' mesons , which we dress by evaluating the corresponding one - loop ( @xmath52-@xmath53 ) selfenergy .
parameters of the model are the resonance mass ( which we fix at 0.5gev above the @xmath52-@xmath54 threshold ) and coupling constant , @xmath55 , which determines the resonance width .
hq and chiral symmetry ( above @xmath33 ) imply degeneracy of the pseudo-/scalar and axial-/vector channels represented by the dirac matrices , @xmath56 .
the temperature dependence of the resulting thermal relaxation times , @xmath57 , for @xmath11- and @xmath18-quarks shows a factor of @xmath83 reduction compared to calculations where only pqcd elastic scattering is accounted for ( right panel of fig . [ fig_tau ] ) .
this is very significant for @xmath11-quarks as @xmath58 is now comparable to ( or even below ) the duration of the qgp phase at rhic ( @xmath39@xmath85fm/@xmath11 ) , whereas @xmath59@xmath60@xmath39 still , due to the large @xmath18-quark mass .
the @xmath29- and @xmath15-dependent drag and diffusion coefficients have been implemented into a relativistic langevin simulation@xcite using an ( elliptic ) thermal fireball expansion for ( semi- ) central au - au collisions at rhic as a background medium , where the bulk flow is adjusted to hydrodynamic simulations in accordance with experiment . at the end of the qgp phase ( which terminates in a mixed phase at @xmath33@xmath1180mev ) ,
the @xmath11- and @xmath18-quark output distributions from the langevin simulation are subjected to hadronization into @xmath50- and @xmath51-mesons using the quark - coalescence model of ref .
since the probability for coalescence is proportional to the light - quark distribution functions , it preferentially occurs at lower @xmath6 ; `` left - over '' heavy quarks are hadronized via @xmath61-function fragmentation . the extra contribution to momentum and @xmath16 from the light
quarks _ increases both _ the @xmath62 and @xmath9 of the heavy - meson spectra , relative to a scheme with fragmentation only .
rescattering in the hadronic phase has been neglected .
the resulting @xmath50- and @xmath51-meson spectra are decayed semileptonically resulting in a nuclear suppression factor and elliptic flow which compare reasonably well with recent rhic data@xcite up to @xmath6@xmath45gev ( fig .
[ fig_elec ] ) . at higher momenta , radiative energy loss ( not included here )
is expected to contribute significantly .
another issue in interpreting the semileptonic @xmath19 spectra concerns the relative contributions of charm and bottom .
since bottom quarks are much less affected by rescattering effects , their contribution reduces the effects in both @xmath63 and @xmath16 appreciably . following refs .
@xcite , we have determined the initial charm spectra by reproducing @xmath50- and @xmath64-spectra in d - au collisions@xcite , calculated the pertinent @xmath19 spectra and assigned the missing yield relative to the data@xcite to bottom contributions . in this way , the crossing between charm and bottom contribution in @xmath15-@xmath15 collisions occurs at @xmath6@xmath45gev . due to the strong quenching of the @xmath11-quark spectra in the qgp this
crossing is shifted down to @[email protected] - 3gev in au - au collisions . for charm contributions
only , our nonperturbative rescattering mechanism results in a pertinent electron @xmath63 as low as 0.1 and an elliptic flow as large as 14% .
thus , the severeness of the `` @xmath19 puzzle '' partially hinges on the baseline spectra in more elementary systems ( @xmath15-@xmath15 and d - au ) .
obviously , a direct measurement of @xmath50-mesons in au - au would be very valuable .
heavy quarkonia ( @xmath11-@xmath65 and @xmath18-@xmath66 bound states ) have long been identified as suitable objects to quantitatively investigate the properties of qcd , including color confinement@xcite .
a particularly appealing feature is the applicability of potential models as a nonrelativistic effective - theory approximation to qcd ( cf . ref .
@xcite for a recent review ) . in the vacuum ,
the heavy - quarkonium spectrum is indeed reasonably well described by the `` cornell potential '' , consisting of a coulomb plus linear confining term .
subsequently it was realized that , when immersing quarkonia into a medium of deconfined color charges , debye screening will reduce the binding and eventually lead to the dissolution of the bound state@xcite ; already slightly above @xmath33 , a substantial ( factor @xmath82 - 3 ) reduction in charmonium and bottomonium binding energies , @xmath67 , has been found , together with the possibility that the ( @xmath68-wave ) ground states ( @xmath69 , @xmath31 , @xmath70 ) survive ( well ) above @xmath33 .
modern lqcd ( -based ) calculations qualitatively support this picture : on the one hand , directly extracted charmonium spectral functions ( in quenched approximation ) indicate resonance peaks up to @xmath82@xmath33@xcite . on the other hand ,
( both quenched and unquenched ) lqcd results for the ( color - singlet ) free energy , @xmath71 have been implemented into potential models within a schrdinger@xcite or lippmann - schwinger equation@xcite .
there is an ongoing debate as to which quantity ( @xmath72 , the internal energy @xmath73 , or a linear combination thereof@xcite ) is the most appropriate one to use as a potential .
maximal binding is obtained with @xmath73 , and even in this case , the ground - state charmonium ( bottomonium ) binding energies are reduced to @xmath740.1 - 0.2gev ( 0.3 - 0.6gev ) at @xmath751.5 @xmath33 , compared to @xmath80.6gev ( 1.1gev ) in vacuum . however
, a quantitative description of in - medium modifications of quarkonia requires at least two additional components , i.e. , the effects of parton - induced dissociation reactions and of in - medium hq masses .
the objective is thus to construct an approach that ( i ) is consistent with finite-@xmath29 lqcd and ( ii ) allows for reliable applications to observables in heavy - ion collisions .
only then the original idea@xcite of using quarkonia as a tool to identify ( and characterize ) confinement ( or at least color screening ) may be realized .
the interplay of medium effects becomes more transparent in terms of the charmonium propagator , schematically written as ( nonrelativistic ) ) in this form implies that the @xmath52-@xmath76 potential is strong enough to generate a pole in the scattering amplitude .
a microscopic ( @xmath29-matrix ) approach is discussed below . ]
@xmath77^{-1 } \ . \label{dpsi}\ ] ] it shows that screening ( affecting the binding energy and thus the real part of the inverse propagator ) , and parton - induced dissociation ( governing the width , @xmath78 , of the spectral function , i m @xmath79 ) are not mutually exclusive but have to be taken into account simultaneously@xcite . the inelastic width ( or dissociation rate ) can be expressed as @xmath80 in terms of ( thermal ) parton - distribution functions , @xmath81 ( @xmath15@xmath1@xmath82 ) , and the parton - induced break - up cross section , @xmath83 .
the latter has first been evaluated for gluo - dissociation@xcite , @xmath84 , using coulomb wave - functions for the quarkonia , leading to a @xmath29-dependent @xmath31 lifetime as shown in the left panel of fig .
[ fig_tau - onium ] .
for the vacuum @xmath31 binding energy , @[email protected] , the lifetime is well behaved showing a rather steep decrease with @xmath29 ( dashed line ) . however , if an in - medium binding energy is used as following from screened ( or lqcd ) potentials , the lifetime is initially reduced ( dotted line ) , but if the binding energy drops below @xmath80.1gev , the phase space in the gluo - dissociation reaction strongly shrinks leading to an unphysical increase of the @xmath31 lifetime with @xmath29 .
this artifact is even more pronounced for the less bound excited states ( @xmath85 , @xmath86 )
. this problem has been remedied by introducing the `` quasifree '' dissociation process@xcite , @xmath87 , where a parton knocks out the @xmath52 ( or @xmath76 ) from the bound state . with reasonable values for the strong coupling constant ( @[email protected] )
the resulting charmonium widths are in the @xmath80.1gev range for @[email protected] @xmath33 ( as relevant for rhic ) with a well - behaved @xmath29-dependence .
the order of magnitude of the inelastic width is easily reproduced using a simplified estimate of eq .
( [ gamma ] ) according to @xmath78@xmath23@xmath88 ; with a parton density @xmath89@xmath410@xmath27 ( at @[email protected] ) , a cross section of 1 mb and @xmath90@xmath11/2 one finds @[email protected] . when applied to bottomonia , the same arguments apply@xcite ; the right panel of fig .
[ fig_tau - onium ] illustrates the impact of in - medium binding energies on bottomonium dissociation : in the rhic temperature regime , screening may lead to up to a factor of @xmath820 ( ! ) reduction in the ground - state @xmath70 lifetime , bringing it down to @xmath82 - 5fm/@xmath11 at @[email protected] - 0.4gev .
this has important consequences for bottomonium suppression at rhic ( as elaborated in sec .
[ ssec_phen ] below ) , rendering it a sensitive probe of color screening@xcite .
a key issue for model descriptions of quarkonium spectral functions in the qgp is the implementation of constraints from lqcd .
the standard object computed on the lattice is ( the thermal expectation value of ) a hadronic current - current correlation function in euclidean time , @xmath91 ( and at zero 3-momentum ) , in quantum - number channel , @xmath92@xcite , @xmath93 the connection to the spectral function , @xmath94 , is given by a convolution with a temperature kernel as @xmath95}{\sinh[\omega/2 t ] } \ . \label{ecor2}\ ] ] while euclidean correlators can be computed with good accuracy ( fig .
[ fig_lat ] ) , their limited temporal extent at finite @xmath29 , 0@xmath7@xmath91@xmath71/@xmath29 , renders the extraction of ( minkowski ) spectral functions more difficult . on the contrary ,
model calculations of spectral functions are readily integrated via eq .
( [ ecor2 ] ) and compared to the rather precise euclidean correlators from lqcd @xcite .
to better exhibit the @xmath29-dependence induced by the spectral function , both lattice and model results for @xmath96 are commonly normalized to the so - called `` reconstructed '' correlator , which follows from eq .
( [ ecor2 ] ) by inserting a free spectral function , @xmath97 .
( [ ecor1 ] ) requires the knowledge of the quarkonium spectral function at all ( positive ) energies . a schematic decomposition consists of a bound - state part and a continuum , @xmath98 where , for fixed @xmath92
, @xmath99 runs over all bound states with masses @xmath100 and couplings @xmath101 , and the threshold energy ( @xmath102 ) and function @xmath103 characterize the onset and plateau of the continuum ( usually taken from pqcd ) . in ref .
@xcite , @xmath29-dependent bound - state masses and couplings have been evaluated by solving a schrdinger equation with in - medium potentials ( screened cornell or lqcd internal energy , @xmath73 ) ; the threshold energy has been inferred from the asymptotic ( @xmath104 ) value of the respective potential , @xmath105 ( which monotonously decreases for @xmath106 ) . for the scalar ( @xmath85 ) channel
it has been found that , despite a rather quickly dissolving bound - state contribution , the correlator increases significantly ( consistent with the lattice result in fig .
[ fig_lat ] , right panel ) , due to the decreasing continuum threshold .
the latter also leads to an increasing pseudoscalar correlator , which is not favored by lqcd ( fig .
[ fig_lat ] , left panel ) . in ref .
@xcite the separation of bound - state and continuum parts is improved by employing a @xmath29-matrix approach@xcite for the @xmath52-@xmath76 interaction , @xmath107 for @xmath108 the fourier transformed and partial - wave expanded lqcd internal energy has been used .
spin - spin ( hyperfine ) interactions are neglected implying degeneracy of states with fixed angular momentum ( @xmath68-wave : @xmath32 and @xmath31 , @xmath109-wave : @xmath110 ) .
@xmath111 denotes the intermediate 2-particle propagator including quark selfenergies . the correlation and spectral functions follow from closing the external legs with 3-momentum integrations as @xmath112 where the coefficient @xmath113 depends on the channel @xmath92 . for a fixed @xmath11-quark mass of @xmath114=1.7gev ,
the resulting charmonium spectral functions confirm that bound states are supported in the @xmath68-wave up to @xmath83@xmath33 while dissolved in the @xmath109-wave below 1.5 @xmath33 , see fig .
[ fig_spec ] . in both cases
, however , one finds a large ( nonperturbative ) enhancement of strength in the threshold region .
the corresponding euclidean correlators are displayed in fig .
[ fig_eucl ] , normalized to a reconstructed one using the vacuum bound - state spectrum and a perturbative continuum with onset at the free @xmath115 threshold , @xmath116=@[email protected] .
the @xmath85 correlator ( right panel ) is well above one , both due to a lower in - medium threshold , @xmath118@xmath1@[email protected] , and the nonperturbative rescattering enhancement .
while this is qualitatively consistent with lqcd , the temperature dependence is opposite , indicative for a further threshold decrease with temperature . in the @xmath32 channel ( left panel ) , the significant reduction in binding energy ( due to color screening ) reduces the correlator at large @xmath91 , qualitatively in line with lqcd .
in fact , the latter shows less reduction , leaving room for a @xmath29-dependent reduction of @xmath118 as well .
the scattering equation ( [ tmat ] ) is furthermore well suited to study finite - width effects .
e.g. , when dressing the @xmath11-quarks with an imaginary selfenergy corresponding to @xmath120@xmath150 mev , inducing a charmonium width of @xmath78@xmath4100mev ( as in the left panel of fig . [ fig_tau - onium ] ) , the euclidean correlators vary by no more than 2% .
this indicates that there is rather little sensitivity to phenomenologically relevant values for the quarkonium widths .
the formation of thermalized matter above @xmath33 in ultrarelativistic heavy - ion collisions ( recall the discussion in the introduction ) provides the basis for describing the production systematics of quarkonia in terms of their in - medium properties as discussed in the previous section .
the suppression of the initially produced number of quarkonia , @xmath121 , may be schematically written as proceeding through 3 stages , @xmath122 with the suppression in the hadron gas ( hg ) believed to be small , and the ( `` pre - equilibrium '' ) nuclear absorption ( cross section ) inferred from @xmath15-@xmath123 experiments , the suppression factor directly probes the dissociation rate of @xmath124 s in the qgp .
the observed @xmath31-suppression pattern at sps can indeed be well described using the quasifree dissociation process with in - medium binding energies consistent with lqcd - potential models@xcite . at rhic energy , copious production of @xmath125 pairs ( @xmath126@xmath420 in central @xmath127=200 agev au - au@xcite , compared to 0.2 at sps ) opens the possibility of secondary charmonium production@xcite .
this is , after all , required by detailed balance in the dissociation reaction , @xmath124+@xmath128+@xmath65+@xmath129 , provided the @xmath124 state still exists at the given temperature@xcite .
if , in addition , the @xmath11-quarks are close to thermal equilibrium , one can apply the rate equation , @xmath130 \quad,\ ] ] for the time evolution of the @xmath31 number .
pertinent predictions ( upper solid line in the left panel of fig .
[ fig_rhic ] ) agree reasonably well with current phenix data@xcite suggesting that the regeneration component becomes substantial in central au - au .
corrections due to incomplete @xmath11-quark thermalization , as well as a lower dissociation ( and thus formation ) temperature , reduce the regeneration ( lower solid curves in the left panel of fig . [ fig_rhic ] ) .
it has also been suggested@xcite that available @xmath31 data at sps and rhic are compatible with suppression of only @xmath85 and @xmath131 ( which make up @xmath840% of the inclusive @xmath31 yield in @xmath15-@xmath15 ) , with the direct @xmath31 s being unaffected .
however , if at rhic the @xmath31 width is of order 0.1gev ( cf .
previous section ) and the qgp lifetime at least 2fm / c , one finds a qgp suppression factor of @[email protected]
. finally , turning to @xmath133 production , the increase of the @xmath133 dissociation rate due to color screening ( recall right panel of fig .
[ fig_tau - onium ] ) could lead to a @xmath133 yield in central au - au ( right panel of fig .
[ fig_rhic ] ) which is more suppressed for the @xmath31 ( left panel ) , and thus provide a rather striking qgp signature@xcite .
charm and bottom hadrons are valuable sensors of the sqgp . in the open - flavor sector , the inadequacy of pqcd to account for the suppression and collectivity of current @xmath19 spectra at rhic may be a rather direct indication of nonperturbative rescattering processes above @xmath33 .
we have elaborated these in a concrete example using ( elastic ) resonance interactions .
the latter will have to be ( i ) evaluated more microscopically , e.g. , using input interactions from lqcd , and ( ii ) combined with radiative energy loss . in the quarkonium sector ,
steady progress is made in implementing first - principle information from finite-@xmath29 lattice qcd into effective models suitable for tests in heavy - ion reactions .
@xmath29-matrix approaches incorporate effects of color - screening , parton - induced dissociation , and in - medium masses and widths of heavy quarks ( which , in turn , connect to the open - flavor sector ) .
we have suggested an sqgp signature in terms of stronger suppression for @xmath133 relative to @xmath31 , which would be a direct proof of @xmath31 regeneration and carries large sensitivity to color screening of bottomonia .
this work was supported in part by a fellowship of the spanish m. e. c. , a feodor - lynen fellowship of the a .- v .-
humboldt foundation , and a u.s .
national science foundation career award under grant phy-0449489 .
adler _ et al .
_ [ phenix collaboration ] , _ phys .
lett . _ * 96 * , 032301 ( 2006 ) .
abelev _ et al . _ [ star collaboration ] , arxiv : nucl - ex/0607012 . h. van hees and r. rapp , _ phys .
c _ * 71 * , 034907 ( 2005 ) .
f. karsch , m.t .
mehr and h. satz , _ z. phys .
c _ * 37 * , 617 ( 1988 ) .
m. asakawa and t. hatsuda , _ phys .
lett . _ * 92 * , 012001 ( 2004 ) .
s. datta _ et al .
_ , _ phys . rev
. d _ * 69 * , 094507 ( 2004 ) .
shuryak and i. zahed , _ phys .
c _ * 70 * , 021901 ( 2004 )
. c.y .
wong , _ phys .
c _ * 72 * , 034906 ( 2005 ) .
alberico _ et al .
_ , _ phys .
d _ * 72 * , 114011 ( 2005 ) .
a. mocsy and p. petreczky , _ phys .
d _ * 73 * , 074007 ( 2006 ) .
m. mannarelli and r. rapp , _ phys .
c _ * 72 * , 064905 ( 2005 ) .
wong , arxiv : hep - ph/0606200 .
g. rpke , d. blaschke and h. schulz , _ phys .
d _ * 38 * , 3589 ( 1988 ) .
shuryak , _ sov .
_ * 28 * , 408 ( 1978 ) ; + g. bhanot and m. peskin , _ nucl . phys . _ * b156 * , 391 ( 1979 ) .
celenza , b. huang and c.m .
shakin , _ phys .
c _ * 59 * , 1030 ( 1999 ) .
l. grandchamp _ et al .
_ , _ phys .
lett . _ * 92 * , 212301 ( 2004 ) . s.s .
adler _ et al .
_ [ phenix collaboration ] , _ phys .
lett . _ * 94 * , 082301 ( 2005 ) .
braun - munzinger and j. stachel , _ phys .
b _ * 490 * , 196 ( 2000 ) .
l. yan , p. zhuang and n. xu , arxiv : nucl - th/0608010 .
h. pereira da costa _ et al .
_ [ phenix collaboration ] , arxiv : nucl - ex/0510051 .
f. karsch , d. kharzeev and h. satz , _ phys .
b _ * 637 * , 75 ( 2006 ) . x. zhao and r. rapp , work in progress ( 2006 ) . | in the first part of the talk , we briefly review the problem of parton - energy loss and thermalization at the relativistic heavy - ion collider and discuss how heavy quarks ( charm and bottom ) can help to resolve the existing experimental and theoretical puzzles .
the second part of the talk is devoted to the properties of heavy quarkonia in the ( strongly interacting ) quark - gluon plasma ( sqgp ) and their consequences for observables in heavy - ion collisions . |
let @xmath1 be a random vector of dimension @xmath2 for some integer @xmath3 , and @xmath4 be its observed value .
our argument is based on the multivariate normal model with unknown mean vector @xmath5 and covariance identity @xmath6 , @xmath7 where the probability with respect to ( [ eq : ynorm ] ) will be denoted as @xmath8 .
let @xmath9 be an arbitrary - shaped region .
the subject of this paper is to compute measures of confidence for testing the null hypothesis @xmath10 . observing @xmath11 ,
we compute a frequentist @xmath0-value , denoted @xmath12 , and also a bayesian posterior probability @xmath13 with a noninformative prior density @xmath14 of @xmath15 .
this is the _ problem of regions _ discussed in literature ; @xcite , @xcite , and @xcite .
the confidence measures were calculated by the bootstrap methods for complicated application problems such as the variable selection of regression analysis and phylogenetic tree selection of molecular evolution .
these model selection problems are motivating applications for the issues discussed in this paper , and the normal model of ( [ eq : ynorm ] ) is a simplification of reality .
let @xmath16 be a sample of size @xmath17 in application problems .
we assume there exists a transformation , depending on @xmath17 , from @xmath18 to @xmath11 so that @xmath19 is approximately normalized .
we assume only the existence of such a transformation , and do not have to consider its details .
since we work only on the transformed variable @xmath19 in this paper for developing the theory , readers may refer to the literature above for the examples of applications . before the problem formulation is given in section [ sec : formulation ] , our methodology is illustrated in simple examples below in this section .
the simplest example of @xmath20 would be the half space of @xmath21 , @xmath22 where the notation @xmath23 , instead of @xmath20 , is used to distinguish this case from another example given in ( [ eq : h0s3 ] ) .
only @xmath24 is involved in this @xmath23 , and one - dimensional normal model @xmath25 is considered .
taking @xmath26 as an alternative hypothesis and denoting the cumulative distribution function of the standard normal as @xmath27 with density @xmath28 , the unbiased frequentist @xmath0-value is given as @xmath29 .
a slightly complex example of @xmath20 is @xmath30 for @xmath31 .
the rejection regions are @xmath32 and @xmath33 with a critical constant @xmath34 , which is obtained as a solution of the equation @xmath35 for a specified significance level @xmath36 .
the left hand side of ( [ eq : defc ] ) is the rejection probability @xmath37 when @xmath15 is on the boundary of @xmath20 , i.e. , @xmath38 or @xmath39 .
the frequentist @xmath0-value is defined as the infimum of @xmath40 such that @xmath20 can be rejected .
this becomes @xmath41 for @xmath42 and @xmath43 for @xmath44 .
considering the case , say , @xmath45 we obtain @xmath46 and @xmath47 .
these two simple cases of @xmath20 and @xmath23 exhibit what @xcite called _ paradox _ of frequentist @xmath0-values .
our simple examples of ( [ eq : h0s2 ] ) and ( [ eq : h0s3 ] ) suffice for this purpose , although they had actually used the spherical shell example explained later in section [ sec : three ] .
@xcite indicated that a confidence measure should be monotonically increasing in the order of set inclusion of the hypothesis .
noting @xmath48 , therefore , it should be @xmath49 , but it is not .
this kind of `` paradox '' can not occur with bayesian methods , and @xmath50 holds always .
considering the flat prior @xmath51 const , say , the posterior distribution of @xmath15 given @xmath11 becomes @xmath52 and the posterior probabilities for the case ( [ eq : numex1 ] ) are @xmath53 and @xmath54 .
the `` paradox '' of frequentist @xmath0-values may be nothing surprise for a frequentist statistician , but a natural consequence of the fact that @xmath55 is for a one - sided test and @xmath12 is for a two - sided test ; the power of testing is higher , i.e. , @xmath0-values are smaller , for an appropriately formulated one - sided test than a two - sided test . in this paper
, we do not intend to argue the philosophical question of whether to be frequentist or to be bayesian , but discuss only computation of these two confidence measures .
computation of the confidence measures is made by the bootstrap resampling of @xcite .
let @xmath56 be a bootstrap sample of size @xmath57 obtained by resampling with replacement from @xmath18 .
the idea of bootstrap probability , which is introduced first by @xcite to phylogenetic inference , is to generate @xmath58 many times , say @xmath59 , and count the frequency @xmath60 that a hypothesis of interest is supported by the bootstrap samples . the bootstrap probability is computed as @xmath61 .
recalling the transformation to get @xmath11 from @xmath18 , we get @xmath62 by applying the same transformation to @xmath58 . for typical problems ,
the variance of @xmath62 is approximately proportional to the factor @xmath63 as mentioned in @xcite .
although we generate @xmath58 in practice , we only work on @xmath62 in this paper .
more specifically , we formally consider the parametric bootstrap @xmath64 which is analogous to ( [ eq : ynorm ] ) but the scale @xmath65 is introduced for multiscale bootstrap .
the bootstrap probability is defined as @xmath66 where @xmath67 denotes the probability with respect to ( [ eq : yboot ] ) . for computing a crude confidence measure , we set @xmath68 , or @xmath69 in terms of @xmath58 , so that the distribution ( [ eq : yboot ] ) for @xmath62 is equivalent to the posterior ( [ eq : postmuflat ] ) for @xmath15 .
this gives an interpretation of the bootstrap probability that @xmath70 for any @xmath20 under the flat prior . in the multiscale bootstrap of @xcite ,
however , we may intentionally alter the scale from @xmath68 , or to change @xmath57 from @xmath17 in terms of @xmath58 for computing @xmath12 .
let @xmath71 be @xmath72 different values of scale , which we specify in advance . in our numerical examples ,
@xmath73 scales are equally spaced in log - scale between @xmath74 and @xmath75 . for each @xmath76 , we generate @xmath58 with scale @xmath77 for @xmath78 times , and observe the frequency @xmath79 .
the observed bootstrap probability is @xmath80 .
how can we use the observed @xmath81 for computing @xmath12 ?
let us assume that @xmath20 can be expressed as ( [ eq : h0s3 ] ) but we are unable to observe the values of @xmath82 and @xmath83 . nevertheless , by fitting the model @xmath84 to the observed @xmath85 , we may compute an estimate @xmath86 of the parameter vector @xmath87 with constraints @xmath31 and @xmath88 .
the confidence measures are then computed as @xmath89 and @xmath90 . in case
we are not sure which of ( [ eq : h0s2 ] ) and ( [ eq : h0s3 ] ) is the reality , we may also fit @xmath91 to the observed @xmath92 s and compare the aic values @xcite for model selection . in practice ,
we prepare collection of such models describing the scaling - law of bootstrap probability , and choose the model which minimizes the aic value .
the examples in section [ sec : intro ] were very simple because the boundary surfaces of the regions are flat . in the following sections , we work on generalizations of ( [ eq : h0s2 ] ) and ( [ eq : h0s3 ] ) by allowing curved boundary surfaces . for convenience ,
we denote @xmath93 with @xmath94 and @xmath95 .
similarly , we denote @xmath96 with @xmath97 . as shown in fig .
[ fig : h03 ] , we consider the region of the form @xmath98 , where @xmath99 and @xmath100 are arbitrary functions of @xmath101 .
this region will reduce to ( [ eq : h0s3 ] ) if @xmath102 for all @xmath101 .
the region may be abbreviated as @xmath103 two other regions @xmath104 and @xmath105 as well as two boundary surfaces @xmath106 and @xmath107 are also shown in fig . [
fig : h03 ] .
we define @xmath108 , or equivalently as @xmath109 the boundary surfaces of the hypotheses are @xmath110 for the region @xmath20 , and @xmath111 for the region @xmath112 .
we do not have to specify the functional forms of @xmath113 and @xmath114 for our theory , but assume that the magnitude of @xmath113 and @xmath114 is very small
. technically speaking , @xmath113 and @xmath114 are _ nearly flat _ in the sense of @xcite . introducing an artificial parameter @xmath115 , a function @xmath116 is called nearly flat when @xmath117 and @xmath118-norms of @xmath116 and its fourier transform are bounded .
we develop asymptotic theory as @xmath119 , which is analogous to @xmath120 with the relation @xmath121 .
the whole parameter space is partitioned into two regions as @xmath122 or three regions as @xmath123 .
these partitions are treated as disjoint in this paper by ignoring measure - zero sets such as @xmath124 .
bootstrap methods for computing frequentist confidence measures are well developed in the literature as reviewed in section [ sec : two ] .
the main contribution of our paper is then given in section [ sec : three ] for the case of three regions . in section [ sec : bayes ] , this new computation method is used also for bayesian measures of @xcite . note that the flat prior @xmath51 const in the previous section was in fact carefully chosen so that @xmath125 for ( [ eq : h0s2 ] ) . this same @xmath14 led to @xmath126 for ( [ eq : h0s3 ] ) .
our definition of @xmath20 given in ( [ eq : h03 ] ) is a simplest formulation , yet with a reasonable generality for applications , to observe such an interesting difference between the two confidence measures .
multiscale bootstrap computation of the confidence measures for the three regions case is described in section [ sec : model ] .
simulation study and some discussions are given in section [ sec : simulation ] and [ sec : discussion ] , respectively .
mathematical proofs are mostly given in appendix .
is the shaded area between surfaces @xmath127 and @xmath128.,scaledwidth=75.0% ]
in this section , we review the multiscale bootstrap of @xcite for computing a frequentist @xmath0-value of `` one - sided '' test of @xmath23 . let @xmath129 be the inverse function of @xmath130 .
the bootstrap @xmath131-value of @xmath23 , defined as @xmath132 , is convenient to work with . by multiplying @xmath65 to it ,
@xmath133 is called the normalized bootstrap @xmath131-value .
theorem 1 of @xcite , as reproduced below , states that the @xmath131-value of @xmath55 is obtained by extrapolating the normalized bootstrap @xmath131-value to @xmath134 , or equivalently @xmath135 in terms of @xmath58 .
[ thm : onesided ] let @xmath23 be a region of ( [ eq : h02 ] ) with nearly flat @xmath113 . given @xmath23 and @xmath11 ,
consider the normalized bootstrap @xmath131-value as a function of @xmath136 ; we denote it by @xmath137 .
let us define a frequentist @xmath0-value as @xmath138 and assume that the right hand side exists .
then for @xmath139 and @xmath36 , @xmath140 meaning that the coverage error , i.e. , the difference between the rejection probability and @xmath40 , vanishes asymptotically as @xmath141 , and that the @xmath0-value , or the associated hypothesis testing , is `` similar on the boundary '' asymptotically .
[ [ proof ] ] proof + + + + + here we show only an outline of the proof by allowing the coverage error of @xmath142 , instead of @xmath143 , in ( [ eq : eh02 ] ) .
this is a brief summary of the argument given in @xcite .
first define the expectation operator @xmath144 for a nearly flat function @xmath116 as @xmath145 where @xmath146 on the right hand side denotes the expectation with respect to ( [ eq : yboot ] ) , that is , for @xmath147 with @xmath148 using the expectation operator , we next define two quantities @xmath149 and work on the bootstrap probability as @xmath150 the third equation is obtained by the taylor series around @xmath151 , and the last equation is obtained by @xmath152 . rearranging ( [ eq : bph02 ] ) , we then get the scaling - law of the normalized bootstrap @xmath131-value as @xmath153 on the other hand , eq .
( 5.10 ) of @xcite shows , by utilizing fourier transforms of surfaces , that ( [ eq : eh02 ] ) holds with coverage error @xmath142 for a @xmath0-value defined as @xmath154 the proof completes by combining ( [ eq : scalelawh02 ] ) and ( [ eq : phinvh02 ] ) . a hypothesis testing is to reject @xmath23 when observing @xmath155 for a specified significance level , say , @xmath156 , and otherwise not to reject @xmath23 .
the left hand side of ( [ eq : eh02 ] ) is the rejection probability , which should be @xmath157 for @xmath158 and @xmath159 for @xmath160 to claim the unbiasedness of the test . on the other hand ,
the test is called similar on the boundary when the rejection probability is equal to @xmath40 for @xmath161 . in this paper
, we implicitly assume that @xmath55 is decreasing as @xmath11 moves away from @xmath23 .
the rejection probability increases continuously as @xmath15 moves away from @xmath23 .
this assumption is justified when @xmath115 is sufficiently small so that the behavior of @xmath55 is not very different from that for ( [ eq : h0s2 ] ) .
therefore , ( [ eq : eh02 ] ) implies that the @xmath0-value is approximately unbiased asymptotically as @xmath119 .
we can think of a procedure for calculating @xmath162 based on ( [ eq : ph02 ] ) . in the procedure ,
the functional form of @xmath163 should be estimated from the observed @xmath92 s using parametric models .
then an approximately unbiased @xmath0-value is computed by extrapolating @xmath163 to @xmath134 .
our procedure works fine for the particular @xmath23 of ( [ eq : h0s2 ] ) , because @xmath164 and @xmath165 .
our procedure works fine also for any @xmath23 of ( [ eq : h02 ] ) when the boundary surface @xmath166 is smooth .
the model is given as @xmath167 using parameters @xmath168 , and thus an approximately unbiased @xmath0-value can be computed by @xmath169 .
it may be interesting to know that the parameters are interpreted as geometric quantities ; @xmath170 is the distance from @xmath11 to the surface @xmath171 , @xmath172 is the mean curvature of the surface , and @xmath173 , @xmath174 , is related to @xmath175-th derivatives of @xmath113 .
however , the series expansion above does not converge , i.e. , @xmath176 does not exist , when @xmath166 is nonsmooth .
for example , @xmath177 serves as a good approximating model for cone - shaped @xmath23 , for which @xmath176 does not take a value of @xmath178 .
this observation agrees with the fact that an unbiased test does not exist for cone - shaped @xmath23 as indicated in the argument of @xcite . instead of ( [ eq : ph02 ] ) ,
the modified procedure of @xcite calculates a @xmath0-value defined as @xmath179 for an integer @xmath180 and a real number @xmath181 .
this is to extrapolate @xmath163 back to @xmath134 by using the first @xmath182 terms of the taylor series around @xmath183 . the coverage error in ( [ eq : eh02 ] )
should reduce as @xmath182 increases , but then the rejection region violates the desired property called monotonicity in the sense of @xcite and @xcite . for taking the balance , we chose @xmath184 and @xmath185 for numerical examples in this paper .
the following theorem is our main result for computing a frequentist @xmath0-value of `` two - sided '' test of @xmath20 . the proof is given in appendix [ app : proof - twosided ] .
[ thm : twosided ] let @xmath20 be a region of ( [ eq : h03 ] ) with nearly flat @xmath113 and @xmath114 . given @xmath20 and @xmath11 ,
consider the approximately unbiased @xmath0-value @xmath186 by applying theorem [ thm : onesided ] to @xmath187 for @xmath188 .
assuming these two @xmath0-values exist , let us define a frequentist @xmath0-value of @xmath20 as @xmath189 for example , ( [ eq : ph03 ] ) holds for the exact @xmath0-value of ( [ eq : h0s3 ] ) defined in section [ sec : intro ] .
then for @xmath190 and @xmath36 , @xmath191 meaning that @xmath12 is approximately unbiased asymptotically as @xmath141 . for illustrating the methodology ,
let us work on the spherical shell example of @xcite , for which we can still compute the exact @xmath0-values to verify our methods .
the region of interest is @xmath192 as shown in panel ( a ) of fig .
[ fig : exregions ] .
we consider the case , say , @xmath193 so that this region is analogous to ( [ eq : numex1 ] ) except for the curvature .
the exact @xmath0-value for @xmath194 is easily calculated knowing that @xmath195 is distributed as the chi - square distribution with degrees of freedom @xmath2 and noncentrality @xmath196 .
writing this random variable as @xmath197 , the exact @xmath0-value is @xmath198 , that is , the probability of observing @xmath199 for @xmath200 .
similarly , the exact @xmath0-value for @xmath201 is @xmath202 . in a similar way as for ( [ eq : h0s3 ] ) , the exact @xmath0-value for @xmath20 is computed numerically as @xmath203 , although the procedure is a bit complicated as explained below .
we first consider the critical constants @xmath204 and @xmath205 for the rejection regions @xmath206 and @xmath207 . by equating the rejection probability to @xmath40 for @xmath208 ,
that is , @xmath209 for @xmath188 , we may get the solution numerically as @xmath210 and @xmath211 for @xmath156 , say .
the @xmath0-value is defined as the infimum of @xmath40 such that @xmath20 can be rejected . to check if theorem [ thm : twosided ] is ever usable , we first compute ( [ eq : ph03 ] ) using the exact values of @xmath212 and @xmath213
. then we get @xmath214 , which agrees extremely well to the exact @xmath215 .
the spherical shell is approximated by ( [ eq : h03 ] ) only locally in a neighborhood of @xmath11 but not as a whole .
nevertheless , theorem [ thm : twosided ] worked fine .
we next think of the situation that bootstrap probabilities of @xmath216 and @xmath217 are available but not their exact @xmath0-values .
we apply the procedure of section [ sec : two ] separately to the two regions for calculating the approximately unbiased @xmath0-values . to work on the procedure ,
here we consider a simple model @xmath218 with parameters @xmath219 for @xmath220 let @xmath221 be the normalized bootstrap @xmath131-value of @xmath187 for @xmath188 . by assuming the simple model for @xmath221 , we fit @xmath222 to the observed multiscale bootstrap probabilities of @xmath187 for estimating the parameters . the actual estimation was done using the method described in section [ sec : highjointbp ] , but we would like to forget the details for the moment .
we get @xmath223 , @xmath224 for @xmath216 , and similarly @xmath225 , @xmath226 for @xmath217 .
@xmath170 s are interpreted as the distances from @xmath11 to the boundary surfaces , and the estimates agree well to the exact values @xmath227 for @xmath216 and @xmath228 for @xmath217
. then the approximately unbiased @xmath0-values are computed by ( [ eq : ph02 ] ) as @xmath229 and @xmath230 , and thus ( [ eq : ph03 ] ) gives @xmath231 , which again agrees well to the exact @xmath215 .
we finally think of a more practical situation , where the bootstrap probabilities are not available for @xmath216 and @xmath217 , but only for @xmath20 .
this situation is plausible in applications where many regions are involved and we are not sure which of them can be treated as @xmath216 or @xmath217 in a neighborhood of @xmath11 ; see @xcite for an illustration .
we consider a simple model @xmath232 , @xmath233 with parameters @xmath234 for @xmath235 by assuming that the two surfaces are curved in the same direction with the same magnitude of curvature @xmath236 . for estimating @xmath237 , ( [ eq : fh0 ] ) is fitted to the observed multiscale bootstrap probabilities of @xmath20 with constraints @xmath238 and @xmath31 , and @xmath86 is obtained as @xmath239 , @xmath240 , @xmath241
. then the approximately unbiased @xmath0-values are computed by ( [ eq : ph02 ] ) as @xmath242 and @xmath243 and thus ( [ eq : ph03 ] ) gives @xmath244 .
this is not very close to the exact @xmath215 , partly because the model is too simple .
however , it is a great improvement over @xmath245 . )
choosing a good prior density is essential for bayesian inference .
we consider a version of noninformative prior for making the posterior probability acquire frequentist properties .
first note that the sum of bootstrap probabilities of disjoint partitions of the whole parameter space is always 1 .
for the two regions case , @xmath246 , and thus @xmath247 .
therefore @xmath248 for the approximately unbiased @xmath0-values computed by ( [ eq : ph02 ] ) , suggesting that we may think of a prior so that @xmath249 .
this was the idea of @xcite to define a bayesian measure of confidence of @xmath20 .
since each of @xmath216 and @xmath217 can be treated as @xmath23 by changing the coordinates , we may assume a prior satisfying @xmath250 it follows from @xmath251 that @xmath252 priors satisfying ( [ eq : pmatch2 ] ) are called probability matching priors .
the theory has been developed in literature @xcite for posterior quantiles of a single parameter of interest .
the examples are the flat prior @xmath51 const for the flat boundary case in section [ sec : intro ] , and @xmath253 for the spherical shell case in section [ sec : three ] .
our multiscale bootstrap method provides a new computation to @xmath254
. we may simply compute ( [ eq : pp03 ] ) with the @xmath212 and @xmath213 used for computing @xmath12 of ( [ eq : ph03 ] ) .
although we implicitly assumed the matching prior , we do not have to know the functional form of @xmath14 .
for the spherical shell example , we may use the exact @xmath212 and @xmath213 to get @xmath255 , or more practically , use only bootstrap probabilities of @xmath20 to get @xmath256 .
we first recall the estimation procedure of @xcite before describing our new proposals for improving the estimation accuracy in the following sections .
let @xmath257 be a parametric model of bootstrap probability such as ( [ eq : fh1 ] ) for @xmath23 or ( [ eq : fh0 ] ) for @xmath20 . as already mentioned in section [ sec : intro ] , the model is fitted to the observed @xmath258 , @xmath259 .
since @xmath79 is distributed as binomial with probability @xmath260 and @xmath78 trials , the log - likelihood function is @xmath261 .
the maximum likelihood estimate @xmath86 is computed numerically for each model .
let @xmath262 denote the number of parameters .
then @xmath263 may be compared for selecting a best model among several candidate models .
@xcite has devised the multistep - multiscale bootstrap as a generalization of the multiscale bootstrap .
the usual multiscale bootstrap is a special case called as the one - step multiscale bootstrap .
our new proposal here is to utilize the two - step multiscale bootstrap for improving the estimation accuracy of @xmath237 , although the two - step method was originally used for replacing the normal model of ( [ eq : ynorm ] ) with the exponential family of distributions .
recalling that @xmath58 is obtained by resampling from @xmath18 , we may resample again from @xmath58 , instead of @xmath18 , to get a bootstrap sample of size @xmath264 , and denote it as @xmath265 .
we formally consider the parametric bootstrap @xmath266 where @xmath267 is a new scale defined by @xmath268 . in @xcite , only the marginal distribution @xmath269 is considered to detect the nonnormality .
for the second step , @xmath270 should have the same functional form as @xmath271 for the normal model .
here we also consider the joint distribution of @xmath272 given @xmath11 .
it is @xmath273-dimensional multivariate normal with @xmath274 .
we denote the probability and the expectation by @xmath275 and @xmath276 , respectively .
then , the joint bootstrap probability is defined as @xmath277 let @xmath278 be a parametric model of @xmath279 or @xmath280 . to work on specific forms of @xmath278 , we need some notations .
let @xmath281 be distributed as bivariate normal with mean @xmath282 , variance @xmath283 , and covariance @xmath284 .
the distribution function is denoted as @xmath285 , where the joint density is explicitly given as @xmath286 .
we also define @xmath287
. then a generalization of ( [ eq : scalelawh02 ] ) is given as follows .
the proof is in appendix [ app : proof - jointbp ] .
[ lem : jointbp ] for sufficiently small @xmath115 , the joint bootstrap probabilities for @xmath23 and @xmath20 are expressed asymptotically as @xmath288 where @xmath289 , @xmath290 , @xmath291 , @xmath292 , and @xmath293 .
thus @xmath278 is specified for @xmath23 as ( [ eq : gh1 ] ) with @xmath294 , @xmath295 using the @xmath296 function of ( [ eq : fh1 ] ) .
similarly , @xmath278 is specified for @xmath20 as ( [ eq : gh0 ] ) with @xmath297 , @xmath298 , @xmath299 , @xmath300 using @xmath301 and @xmath302 functions of ( [ eq : fh0 ] ) .
we may specify @xmath72 sets of @xmath303 , denoted as @xmath304 . in our numerical examples ,
@xmath305 are specified as mentioned in section [ sec : intro ] and @xmath306 s are specified so that @xmath307 holds always , meaning @xmath308 . for each @xmath259 , we generate @xmath272 with @xmath309 many times , say @xmath310 , and observe the frequencies @xmath311 , @xmath312 , and @xmath313 .
note that only one @xmath314 is generated from each @xmath62 here , whereas thousands of @xmath314 s may be generated from each @xmath62 in the double bootstrap method .
the log - likelihood function becomes @xmath315 .
in fact , we have used this two - step multiscale bootstrap , instead of the one - step method , in all the numerical examples .
the one - step method had difficulty in distinguishing @xmath20 with very small @xmath83 from @xmath20 with moderate @xmath83 but heavily curved @xmath127 .
the two - step method avoids this identifiability issue because a small value of @xmath316 indicates that @xmath83 is small ; it is automatically done , of course , by the numerical optimization of @xmath317 .
the asymptotic errors of the scaling law of the bootstrap probabilities in ( [ eq : bph02 ] ) and ( [ eq : gh1 ] ) are of order @xmath142 .
as shown in the following lemma , the errors can be reduced to @xmath143 by introducing correction terms of @xmath142 for improving the parametric model @xmath278 of @xmath23 . the proof is given in appendix [ app : proof - jointbph ] .
[ lem : jointbph ] for sufficiently small @xmath115 , the bootstrap probabilities for @xmath23 are expressed asymptotically as @xmath318 where @xmath151 , @xmath319 , and @xmath320 are those defined in lemma [ lem : jointbp ] , and the higher order correction terms are defined as @xmath321 , @xmath322 , and @xmath323 using @xmath324 for deriving a very simple model for @xmath325 , we think of a situation @xmath326 and @xmath327 , and consider asymptotics as @xmath328 .
this formulation is only for convenience of derivation .
the two values @xmath329 and @xmath59 will be specified later by looking at the functional form of @xmath257
. a straightforward , yet tedious , calculation ( the details are not shown ) gives @xmath330 and @xmath331 this correction term was in fact already used for the simple model @xmath218 of the spherical shell example in section [ sec : three ] , where the parameter was actually @xmath332 instead of @xmath333 .
we did not change the @xmath163 for adjusting @xmath334 and @xmath335 , meaning that @xmath336 , instead of @xmath151 , was modelled as @xmath337 . comparing the coefficients of @xmath163 , we get @xmath338 and @xmath339 , and thus @xmath340 . when ( [ eq : fh1 ] ) was fitted to @xmath216 , the estimated parameter @xmath341 was close to the true value @xmath342 .
for the numerical example mentioned above , we have also fitted the same model but @xmath343 being fixed .
the estimated parameters are @xmath344 , @xmath345 , and the @xmath0-value is @xmath346 .
these values are not much different from those shown in section [ sec : three ] . however , the aic value improved greatly by the introduction of @xmath325 , and the aic difference was 96.67 , mostly because improved fitting for the joint bootstrap probability of ( [ eq : achjoint ] ) .
my experience suggests that consideration of the @xmath325 term is useful for choosing a reasonable model of @xmath163 .
let us consider a cone - shaped region @xmath20 in @xmath347 with the angle at the vertex being @xmath348 as shown in panel ( b ) of fig .
[ fig : exregions ] .
this cone can be regarded , locally in a neighborhood of @xmath11 with appropriate coordinates , as @xmath20 of ( [ eq : h03 ] ) when @xmath11 is close to one of the edges but far from the vertex , or as @xmath23 of ( [ eq : h02 ] ) when @xmath11 is close to the vertex . in this section , the cone
is labelled either by @xmath20 or @xmath23 depending on which view we are taking .
cones in @xmath347 appear in the problem of multiple comparisons of three elements @xmath349 , say , and @xmath187 corresponds to the hypothesis that the mean of @xmath350 is the largest among the three @xcite .
the angle at the vertex is related to the covariance structure of the elements .
although an unbiased test does not exist for this region , we would like to see how our methods work for reducing the coverage error .
contour lines of confidence measures , denoted @xmath351 in general , at the levels 0.05 and 0.95 are drawn in fig .
[ fig : rejregion10 ] .
the rejection regions of the cone and the complement of the cone are @xmath352 and @xmath353 , respectively , at @xmath156 .
we observe that @xmath351 decreases as @xmath11 moves away from the cone in panels ( a ) , ( b ) , and ( c ) ; see appendix [ app : sim ] for the details of computation . on the other hand , figs .
[ fig : rejprob10 ] and [ fig : rejprob20 ] show the rejection probability . for an unbiased test
, it should be 5% for all the @xmath354 so that the coverage error is zero . in panel ( a ) of fig .
[ fig : rejregion10 ] , @xmath355 is computed by the bootstrap samples of @xmath356 .
this bootstrap probability , labelled as bp in fig .
[ fig : rejprob10 ] , is heavily biased near the vertex , and this tendency is enhanced when the angle becomes @xmath357 in fig .
[ fig : rejprob20 ] . in panel ( b ) of fig .
[ fig : rejregion10 ] , @xmath358 is computed by regarding the cone as @xmath23 of ( [ eq : h02 ] ) .
the dent of @xmath359 and the bump of @xmath360 become larger than those of panel ( a ) of fig .
[ fig : rejregion10 ] near the vertex , confirming what we observed in @xcite . as seen in figs .
[ fig : rejprob10 ] and [ fig : rejprob20 ] , the coverage error of @xmath55 , labelled as `` one sided '' there , is smaller than that of bp . in panel ( c ) of fig .
[ fig : rejregion10 ] , @xmath12 is also computed by regarding the cone as @xmath20 of ( [ eq : h03 ] ) , and then one of @xmath55 and @xmath12 is selected as @xmath351 by comparing the aic values at each @xmath11 .
this @xmath351 , labelled as `` two sided freq '' in figs .
[ fig : rejprob10 ] and [ fig : rejprob20 ] , improves greatly on the one - sided @xmath0-value .
the coverage error is almost zero except for small @xmath361 s , verifying what we attempted in this paper .
the corresponding bayesian posterior probability , labelled as `` two sided bayes , '' performs similarly .
note that the coverage error was further reduced near the vertex by setting simply @xmath362 without the model selection ( the result is not shown here ) ; however , the shapes of @xmath363 and @xmath360 became rather weird then in the sense mentioned at the last paragraph of section [ sec : two ] . and @xmath364 .
the cone - shaped region @xmath20 is rotated so that one of the edges is placed along the x - axis .
solid curves are drawn for ( a ) the bootstrap probability with @xmath356 , and for ( b ) the frequentist @xmath0-value for `` one - sided '' test . in panel ( c ) , @xmath351 is switched to the frequentist @xmath0-value for `` two - sided '' test when appropriate . the dotted curve in panel ( c )
is for the bayesian posterior probability.,scaledwidth=70.0% ] .,scaledwidth=75.0% ] .,scaledwidth=75.0% ]
in this paper , we have discussed frequentist and bayesian measures of confidence for the three regions case , and have proposed a new computation method using the multiscale bootstrap technique . in this method , aic played an important role for choosing appropriate parametric models of the scaling - law of bootstrap probability .
simulation study showed that the proposed frequentist measure performs better for controlling the coverage error than the previously proposed multiscale bootstrap designed only for the two regions case .
a generalization of the confidence measures gives a frequentist interpretation of the bayesian posterior probability as follows .
let us consider the situation of theorem [ thm : twosided ] .
if we strongly believe that @xmath365 , we could use the one - sided @xmath0-value @xmath366 , instead of the two sided @xmath12 .
similarly , we might use @xmath367 if we believe that @xmath368 . by making the choice
`` adaptively , '' someone may want to use @xmath369 , although it is not justified in terms of coverage error . by connecting @xmath370 and @xmath12 linearly using an index @xmath371 for the number of `` sides , '' we get @xmath372
it is easily verified that @xmath373 and @xmath374 , indicating that the bayesian posterior probability defined in section [ sec : bayes ] can be interpreted , interestingly , as a frequentist @xmath0-value of `` zero - sided '' test of @xmath20 .
although we have no further consideration , this kind of argument might lead to yet another compromise between frequentist and bayesian .
our formulation is rather restrictive .
we have considered only the three regions case by introducing the surface @xmath114 in addition to the surface @xmath113 of the two regions case . also these two surfaces are assumed to be nearly parallel to each other .
it is worth to elaborate on generalizations of this formulation in future work , but too much of complication may result in unstable computation for estimating the scaling - law of bootstrap probability .
aic will be useful again in such a situation .
first we consider rejection regions of testing @xmath20 for a specified @xmath40 by modifying the two rejection regions of ( [ eq : h0s3 ] ) .
since @xmath113 and @xmath114 are nearly flat , the modified regions should be expressed as @xmath375 and @xmath376 using nearly flat functions @xmath377 and @xmath378 .
the constant @xmath34 is the same one as defined in ( [ eq : defc ] ) .
write @xmath379 , @xmath380 for brevity sake .
we evaluate the rejection probability for @xmath381 .
let @xmath382 for a moment , and put @xmath383 . by applying the argument of ( [ eq : bph02 ] ) to @xmath384 but ( [ eq : yboot ] )
is replaced by ( [ eq : ynorm ] ) , we get @xmath385 .
the same argument applied to @xmath386 gives @xmath387 . rearranging these two formula with the identity @xmath388 for an unbiased test
, we get an equation @xmath389 . by exchanging the roles of @xmath377 and @xmath378 ,
the equation becomes @xmath390 for @xmath391 with @xmath392 .
these two equations are expressed as @xmath393 for solving this equation with respect to @xmath377 and @xmath378 , first apply the inverse matrix of the @xmath394 matrix from the left in ( [ eq : prftwo - e ] ) , and then apply the inverse operator of @xmath395 so that @xmath396 next we obtain an expression of @xmath0-value corresponding to the rejection regions .
@xmath12 is defined as the value of @xmath40 for which either of @xmath397 and @xmath398 holds .
note that @xmath377 , @xmath378 , and @xmath34 depend on @xmath40 .
let us assume @xmath397 and thus @xmath399 for a moment .
write @xmath400 , @xmath401 for brevity sake . recalling ( [ eq : defc ] ) , @xmath402 , where @xmath403 in ( [ eq : prftwo - r ] ) can be expressed as @xmath404 therefore , @xmath405 . by applying ( [ eq : phinvh02 ] ) to @xmath216 and @xmath217 , respectively , we get @xmath406 and @xmath407 , and thus @xmath408 . by exchanging the roles of @xmath216 and @xmath217
, we have @xmath409 for @xmath410 . by taking the minimum of these two expressions of @xmath12 , we finally obtain ( [ eq : ph03 ] ) .
this @xmath0-value satisfies ( [ eq : prftwo - a ] ) with error @xmath142 , and thus ( [ eq : eh03 ] ) holds .
the argument is very similar to ( [ eq : bph02 ] ) in the proof of theorem [ thm : onesided ] . given @xmath411 , the joint distribution of @xmath412 and @xmath413 is @xmath414 .
therefore , @xmath415 , where @xmath416 and @xmath417 are defined in ( [ eq : e1e2 ] ) .
taking the expectation with respect to @xmath418 , we have @xmath419 . for proving ( [ eq : gh1 ] ) , considering the taylor series around @xmath420
, we obtain @xmath421 with @xmath422 for completing the proof .
next we show ( [ eq : gh0 ] ) . the conditional probability given
@xmath423 is @xmath424 , where @xmath425 taking the expectation with respect to @xmath418 , we have @xmath426 .
we only have to consider the taylor series @xmath427 with @xmath428 , @xmath188 for completing the proof . by considering a higher - order term of the taylor series in ( [ eq : bph02 ] ) , we obtain @xmath429 , proving ( [ eq : ach1 ] ) as well as ( [ eq : ach2 ] ) . on the other hand , ( [ eq : achjoint ] ) is shown by considering higher - order terms of the taylor series in ( [ eq : expjointbp ] ) as @xmath430 the proof completes by rearranging the above formula with @xmath431
the contour lines in fig . [ fig : rejregion10 ] are drawn by computing @xmath0-values at all grid points ( @xmath432 ) of step size 0.05 in the rectangle area ; this huge computation was made possible by parallel processing using up to 700 cpus .
the computation takes a few minutes per each grid point per cpu .
our algorithm is implemented as an experimental version of the scaleboot package of @xcite , which will be included soon in the release version available from cran .
the rejection probabilities in figs .
[ fig : rejprob10 ] and [ fig : rejprob20 ] are computed by generating @xmath11 according to ( [ eq : ynorm ] ) for 10000 times , and then counting how many times @xmath433 or @xmath434 is observed .
this computation is done for each @xmath435 with the distance from the vertex @xmath436 , i.e. , @xmath437 in the coordinates of fig .
[ fig : rejregion10 ] . for computing @xmath438 and @xmath12 , the two - step multiscale bootstrap described in section [ sec : twostep ] was performed with the @xmath73 sets of scales @xmath439 , @xmath440 , specified there .
the parametric bootstrap , instead of the resampling , was used for the simulation .
the number of bootstrap samples has increased to @xmath441 for making the contour lines smoother , while it was @xmath442 in the other results . for @xmath438
, we have considered the singular model of @xcite defined as @xmath443 for cones , and performed the model fitting method described in section [ sec : highjointbp ] . from the taylor series of this @xmath163 around @xmath68 , we get @xmath444 , @xmath445 for computing the higher order correction term @xmath325
. we have also considered submodels by restricting some of @xmath446 to specified values , and the minimum aic model is chosen at each @xmath11 .
the frequentist @xmath0-value is computed by ( [ eq : pk ] ) with @xmath184 and @xmath185 . for @xmath12
, we have considered the same singular model for the two surfaces by assuming they are curved in the opposite directions with the same magnitude of curvature .
more specifically , the two @xmath296 functions in ( [ eq : fh0 ] ) are defined as @xmath447 and @xmath448 .
the parameters @xmath449 are estimated by the model fitting method described in section [ sec : twostep ] .
submodels are also considered and model selection is performed using aic .
the frequentist @xmath0-value is computed by ( [ eq : ph03 ] ) , and the bayesian posterior probability is computed by ( [ eq : pp03 ] ) . the rejection probabilities of other two commonly used measures are shown only for reference purposes ; see @xcite for the details .
the rejection probability of the multiple comparisons , denoted mc here , is always below 5% in panel ( a ) , and the coverage error becomes zero at the vertex . on the other hand ,
the rejection probability of the @xmath131-test is always below 5% in panel ( b ) , and the coverage error reduces to zero as @xmath450 . | a new computation method of frequentist @xmath0-values and bayesian posterior probabilities based on the bootstrap probability is discussed for the multivariate normal model with unknown expectation parameter vector .
the null hypothesis is represented as an arbitrary - shaped region .
we introduce new parametric models for the scaling - law of bootstrap probability so that the multiscale bootstrap method , which was designed for one - sided test , can also computes confidence measures of two - sided test , extending applicability to a wider class of hypotheses .
parameter estimation is improved by the two - step multiscale bootstrap and also by including higher - order terms .
model selection is important not only as a motivating application of our method , but also as an essential ingredient in the method .
a compromise between frequentist and bayesian is attempted by showing that the bayesian posterior probability with an noninformative prior is interpreted as a frequentist @xmath0-value of `` zero - sided '' test . |
quantum - speed - limit ( qsl ) time @xcite , the intrinsic minimal time interval for a quantum system evolving from an initial state to a target state , is of crucial importance in the fields of quantum computation quantum computer , quantum control control1,control2,control3,control4,control5 , quantum metrology metrology , metrology1,metrology2 , and non - equilibrium thermodynamics thermo .
recent decades have witnessed a great deal of research on qsl time both in closed @xcite and open systems @xcite . in particular , a qsl time based on the schatten @xmath1 norm for an arbitrarily driven open system was presented by deffner and lutz @xcite @xmath2}{\min \left\ { e_{1}^{\tau } , e_{2}^{\tau } , e_{\infty } ^{\tau } \right\ } } , \label{qsl}\ ] ] where @xmath3 denotes the bures angle between the initial state @xmath4 and the target state @xmath5 , which is governed by the time - dependent master equation @xmath6 ( @xmath7 is a superoperator ) .
it is reasonable to employ the bures angle above as a measure for distance between an pure state and a mixed state , since the important riemannian feature is satisfied under such a circumstance @xcite .
the denominator @xmath8 in eq .
( [ qsl ] ) represents the average of @xmath9 over actual driving time duration @xmath10 , i.e. , @xmath11 .
one important application of the qsl time is to evaluate the speed of quantum evolutions under the following two scenarios : ( i ) .
the fastest evolution appears when the actual driving time @xmath10 achieves the qsl time @xmath12 , i.e. , @xmath13 for slower evolutions , @xmath14 and the slower the evolution , the higher ratio @xmath15 should be .
this point of view is clear and unambiguous and has been widely utilized in the field of entanglement assisted speedup of quantum evolutions @xcite .
( ii ) . from another point of view
, @xmath16 indicates the evolution is already the fatest , and possesses no potential capacity for further acceleration , while for @xmath17 the higher ratio @xmath18 ( or equivalently , the much shorter @xmath12 ) , the greater the capacity for potential speedup will be .
this viewpoint has been adopted in exploring the memory effect , characterized by non - markovianity @xcite , on the speed of quantum evolution qslopen3,xu1,xu2 .
it is found that in the damped jaynes - cummings model of a single qubit , the transition from no potential capacity of speedup ( @xmath19 ) to potential speedup ( @xmath14 ) , or say a reduction of qsl in quantum evolution is just the critical point when the memoryless environment becomes of memory @xcite .
furthermore , if evolution of a qubit is already along the qsl time which means its fastest speed in nature environment , the study on it will probably not so urgent as the case of @xmath14 which have speedup potential .
although the memory effect of environment plays a decisive role in the potential acceleration of quantum evolution in a single qubit case , the question may arise whether it will still be true even in multi - qubit cases .
this question is of particular interest , for the qsl time of multi - qubit open systems has now caught increasing attention and several interesting phenomena were discovered . for instances , with the qsl time defined in ref .
( @xcite ) , taddei _
et al_. illustrated that the separable states of a multi - qubit system under markovian dephasing channels perform the same speedup of quantum evolution as the entangled states when the number of qubits is large enough .
on the other hand , del campo _ et al . _ demonstrated , with the bound introduced in ref .
( @xcite ) , that multi - qubit greenberger - horne - zeilinger ( ghz ) states and separable states under markovian and non - markovian dephasing are equivalent in metrological parameter estimation @xcite . in this paper , mainly focused on scenario 2 , we show that for a class of multiqubit states memory effect is a requisite for raising potential ability of speeding up quantum evolution , while for another class of multi - qubit states , memoryless environment is ready to realize the above result .
especially when the open system is initially prepared in product state @xmath20 a transition from no potential capacity for speedup to possess speedup ability is achieved .
the paper is organized as follows : in sec .
ii , we present the qsl time for typical two - qubit , three - qubit , and n - qubit states , respectively .
discussion on the role of entanglement in qsl time is performed in sec .
finally , conclusions are drawn in sec . iv .
we consider n independent two - level atoms ( open system ) each locally coupling to a leaky vacuum cavity ( environment ) .
the dynamics of the multi - qubit open system is fully determined by each pair of atom cavities @xcite with the following hamiltonian @xcite @xmath21 where @xmath22 is the resonant transition frequency of the atom between the excited state @xmath23 and the ground state @xmath24 , and @xmath25 are the pauli raising and lowering operators . @xmath26 and @xmath27 denote the frequency and the annihilation(creation ) operators of the @xmath28th mode of the cavity with @xmath29 the corresponding real coupling constant .
the master equation for the reduced density matrix of the atom in the schrdinger picture is given by @xmath30 with @xmath31 + \gamma _ { t}\left ( { \sigma } _ { + } { \sigma } _ { -}\rho _ { t}+\rho _ { t}{\sigma } _ { + } % { \sigma } _ { -}-2{\sigma } _ { -}\rho _ { t}{\sigma } _ { + } \right ) , \label{master equation}\]]where @xmath32=im@xmath33 and @xmath34=re@xmath35 are the time - dependent lamb shift and decay rate respectively , and @xmath36 is the decoherence function relying on the particular structure of cavity reservoirs @xcite . the reduced density matrix of the atom with an initial state @xmath37 takes the form @xmath38 , \label{density matrix}\ ] ] where the excited state population @xmath39 is denoted by @xmath40 in the following .
as the exact form of qsl time for a general pure initial state is cumbersome , we only consider two typical bell - type initial states respectively , i.e. , @xmath41 and @xmath42 with @xmath43 $ ] . according to the definition of qsl time in eq .
( [ qsl ] ) , we have @xmath44for state @xmath45 and @xmath46 for state @xmath47 , where @xmath48 of a two - qubit open system under a memoryless environment as a function of @xmath49 with initial states @xmath50 ( black dotted line ) , @xmath51 ( red solid curve ) , and @xmath52 @xmath53 ( blue dashed curve ) .
the red circle indicates the maximal speedup of quantum evolution for state @xmath54,width=336 ] obviously , the qsl time ratio @xmath12 of state @xmath55 is the same as the exact form in single qubit cases discovered in refs @xcite , where the reduction in qsl time ( @xmath56 ) only occurs when the environment is of memory ; otherwise the evolution will always be along the qsl time ( @xmath57 ) . in order to illustrate this phenomenon clearly , the qsl time ratio @xmath15 of initial state @xmath55 ( black dotted line )
is depicted in fig . 1 under a memoryless environment , i.e. , @xmath58 with the population @xmath40 monotonically decreasing from 1 to the target @xmath59 .
however , a complex but interesting phenomenon appears for initial state @xmath60 : qsl reduction can also take place even when the environment is of no memory . for instances , @xmath15 versus @xmath59 of initial states
@xmath61 ( red solid curve ) and @xmath62 @xmath63 ( blue dashed curve ) are depicted , respectively , in fig .
1 , where @xmath64 clearly illustrates the intrinsic acceleration potential of quantum evolution under a memoryless environment . especially , there exists a sudden change point in qsl time with the critical point @xmath65 for initial state @xmath66 @xmath67 . to explain this phenomenon ,
we trace back to @xmath68 defined in eq .
( [ qsl ] ) with the following expression @xmath69 where we have employed the condition of memoryless environment @xmath70 therefore , the qsl time ratio @xmath15 of eq .
( [ qsl2 - 2 ] ) can be conveniently calculated as @xmath71one may check that @xmath72 is always satisfied .
in particular , when the initial state is in @xmath73 , the qsl time ratio yields @xmath74the sudden change point of qsl time is therefore justified . in this subsection
, we also consider two typical three - qubit states , i.e. , w type state @xmath75for state @xmath76 and @xmath77 for state @xmath78 , where @xmath79 equation ( [ qsl3 - 1 ] ) bears a resemblance to the case of state @xmath80 , implying that the reduction in qsl only occurs in a memory environment . as for eq .
( [ qsl3 - 3 ] ) , we consider a special case , i.e. , @xmath81 under the environment of no memory ( @xmath82 ) . therefore ,
( qsl3 - 3 ) can be simplified as : @xmath83 indicating that there also exists a sudden transition of speedup potential of quantum evolution even the environment is of no memory .
in this subsection , we show that the above phenomena are ubiquitous in n - qubit cases ( n is an arbitrary positive integer ) .
it is easy to check that if the n - qubit open system is initially prepared in state @xmath84 with @xmath85 the qsl ratio is exactly the same as eqs .
( [ qsl2 - 1 ] ) and ( [ qsl3 - 1 ] ) .
therefore , the memory effect of environment becomes the essential condition for the speedup potential emerge in quantum evolution .
however , if the initial state is in @xmath86 , the qsl ratio is given by @xmath87 in particular when the environment is memoryless , i.e. , @xmath88 eq .
( [ qsln-1 ] ) reduces to @xmath89clearly , @xmath65 is the critical point at which the open system experiences the sudden change of qsl time under a memoryless environment . equation ( [ qsln-2 ] ) also implies that there exists a maximal acceleration potential condition for state @xmath90 when @xmath91 , and the corresponding minimal qsl time ratio is given by : @xmath92 especially when @xmath93 , eq .
( [ maximal speedup ] ) grows to @xmath94 , which is marked in fig . 1 as the red circle . of a lorentzian spectral distribution model versus coupling strength @xmath95
the black dotted , red solid , and blue dashed curves represent the two - qubit initial states @xmath96 , @xmath97 , and @xmath98 @xmath99 , respectively , with parameters @xmath100 , @xmath101 , and @xmath102,width=336 ] in this subsection , we intend to show that memory effect of environment is still an important element for quantum acceleration potential for multi - qubit open systems .
the memory environment we consider here is characterized by the lorentzian spectral distribution @xmath103 , where @xmath104 is the markovian decay rate and @xmath105 is the spectral width @xcite .
@xmath40 is now written as @xcite@xmath106where @xmath107 . in fig .
2 , we take two - qubit initial states as an example and fix the driving time as @xmath108the qsl time ratio @xmath15 versus coupling strength @xmath109 is plotted with three typical two - qubit states @xmath110 ( black dotted line ) , @xmath111 ( red solid curve ) , and @xmath112 ( blue dashed curve ) , respectively , with @xmath113 , @xmath114 , and we set @xmath115 . according to ref .
( book open ) , we know that @xmath116 is the transient point from memoryless environment ( @xmath117 ) to one of memory ( @xmath118 ) . as is clearly shown in fig .
2 , more capacity for potential speedup will take place when the environment enters the memory region .
considering that the qsl time for multiqubit systems can be shortened in a memoryless environment , it is natural to link this phenomenon to entanglement , which only exists in multiqubit cases . in closed composite systems
, entanglement has been taken as a resource in the speedup of quantum evolution [ @xcite ] . in this subsection
, we go a step further to the connection between entanglement and qsl time in bipartite open systems .
the initial state we consider here is an arbitrary pure state : @xmath119 with @xmath120 which is generated by monte carlo method , and the related entanglement is characterized by concurrence @xmath121 in ref .
@xcite , with @xmath122 for a disentangled state and @xmath123 for a maximally entangled state . as a distribution of concurrence @xmath121 for 20000 randomly generated initial pure states ( tiny blue dots ) under the damping model .
the dark red and light green dots represent qsl time ratio with states @xmath124 and @xmath125 respectively.,width=336 ] in fig . 3 ,
20000 random pure states are generated by monte carlo sampling @xcite and their qsl ratios @xmath15 and their qsl time ratios @xmath15 under memoryless environments versus concurrence are marked by tiny blue dots . as is clearly displayed in fig .
3 , the lower bound @xmath16 can always be reached by states @xmath126 ( dark red dots in fig .
3 ) , implying that the concurrence has nothing to do with @xmath18 under this circumstance .
in addition , the upper bound is taken by a subset of states @xmath127 conclusion ---------- in summary , we have explored the quantum - speed - limit time for multi - qubit open systems . for a certain class of initial states , we have demonstrated that the quantum evolution can also be accelerated in a memoryless ( markovian ) environment .
moreover , we have found that entanglement plays a subtle role in the speedup of quantum evolution : weak entanglement may be better for speeding up quantum evolution under certain circumstances .
we have only treated non - correlated environments in this paper .
it will also be of importance and interest to study the qsl time of multi - qubit systems with the presence of initial correlations among the subsystems of composite environments @xcite .
this work was supported by the national natural science foundation of china ( grants no .
11204196 and no . 11074184 ) , the specialized research fund for the doctoral program of higher education ( grant no .
20123201120004 ) , and the priority academic program development of jiangsu higher education institutions . | quantum - speed - limit ( qsl ) time captures the intrinsic minimal time interval for a quantum system evolving from an initial state to a target state . in single qubit open systems
, it was found that the memory ( non - markovian ) effect of environment plays an essential role in shortening qsl time or , say , increasing the capacity for potential speedup . in this paper
, we investigate the qsl time for multiqubit open systems .
we find that for a certain class of states the memory effect still acts as the indispensable requirement for cutting the qsl time down , while for another class of states this takes place even when the environment is of no memory .
in particular , when the initial state is in product state @xmath0 , there exists a sudden transition from no capacity for potential speedup to potential speedup in a memoryless environment .
in addition , we also display evidence for the subtle connection between qsl time and entanglement that weak entanglement may shorten qsl time even more . |
the structure and origin of the galactic disk(s ) are among the hottest debating issues of the galactic astronomy .
an archeological approach to the problems relies on the collection of information in multi - dimensional phase space for large samples of stars .
the on - going lamost spectroscopic survey of the galactic anti - center ( lss - gac ) , will collect medium - to - low resolution ( @xmath5 @xmath61800 ) spectra of millions of stars and deliver fundamental stellar parameters , including radial velocity ( @xmath0 ) and stellar atmospheric parameters ( effective temperature @xmath1 , surface gravity log@xmath2 , metallicity [ fe / h ] ) , as well as elemental abundance ratios ( the @xmath7element to iron abundance ratio [ @xmath8/fe ] , and the carbon to iron abundance ratio [ c / fe ] ) , deducible from the spectra ( liu et al .
combined with accurate optical and infrared ( ir ) photometry , the lss - gac will also deliver estimates of the interstellar extinction and distances to individual stars ( yuan et al .
2014 , submitted ; paperiii hereafter ) . together with determinations of proper motions , either from existing catalogs , such as the ppmxl ( roeser et al .
2010 ) and ucac4 ( zacharias et al .
2013 ) , or from the forth coming measurements of gaia of unprecedented accuracy ( perryman et al .
2001 ) , as well as accurate parallaxes ( distances ) also provided by gaia , the lss - gac will provide an unprecedented large stellar database in multi - dimensional phase space to study the stellar populations , kinematics and chemistry of the galactic disk and its assemblage and evolution history .
deriving accurate stellar parameters from millions of medium - to - low resolution fiber spectra in an efficient way is thus of fundamental importance to fulfill the scientific goals of lss - gac .
various methods have been developed in the past to derive stellar atmospheric parameters from large number of medium - to - low resolution spectra ( recio - blanco et al .
2006 ; lee et al .
2008a , wu et al .
the approaches generally fall into two main categories of method ( wu et al .
2011 ) : the minimum distance method ( mdm ) and non - linear regression method . both categories of method have been applied to large stellar spectroscopic surveys , including the segue ( yanny et al .
2009 ) , rave ( steinmetz et al .
2006 ) , apogee ( majewski et al . 2007 ) and lamost ( zhao et al . 2012 ) .
the mdm is usually based on spectral template matching , and searches for the template spectrum that has the shortest distance in parameter space from the target spectrum .
the @xmath9 minimization , cross - correlation , weighted mean algorithm , and the @xmath10-nearest neighbor ( knn ) , are thought to be specific cases of mdm ( wu et al .
softwares and pipelines developed based on those algorithms include the tgmet ( katz et al .
1998 ) , matisse ( recio - blanco et al . 2006 ) , sspp ( lee et al .
2008a ) , ulyss ( koleva et al .
2009 ) , that of allende prieto et al .
( 2006 ) and of zwitter et al .
the non - linear regression method is sometimes also referred to as the artificial neural network ( ann ) .
the method constructs a functional mapping between the spectra and stellar atmospheric parameters by training a library of template spectra with non - linear algorithms such as the principal component analysis ( pca ) , and then apply the mapping to target spectra .
related work can be found in re fiorentin et al .
( 2007 ) and lee et al .
( 2008a ) .
in addition to the above two categories of method , other approaches have been developed , for example , the line - index method based on the relations between the stellar atmospheric parameters and the equivalent widths of spectral features and/or photometric colours ( wilhelm et al .
1999 ; beers et al . 1999 ; cenarro et al .
more recently , a bayesian approach to determine stellar atmospheric parameters combing spectral and photometric measurements has been developed by schnrich & bergemann ( 2013 ) .
different methods usually have different valid parameter ranges , outside which the methods perform poorly .
for example , the line - index method usually loses sensitivity when the adopted metallic lines are either saturated or too weak . because stars are widely distributed in the parameter space and different spectral features have different sensitivity to the atmospheric parameters
, one can hardly rely on one single method to derive stellar atmospheric parameters with a uniform accuracy for all types of star .
a `` multi - method '' approach , which takes averaged stellar atmospheric parameters deduced from a variety of methods that utilizes different spectral wavelength ranges , is adopted by the sspp ( lee et al .
since systematic errors from different methods can not be easily compared and combined , the systematic errors of the final parameters are difficult to estimate .
almost all of the methods determine the stellar atmospheric parameters via either direct or indirect comparisons between the target and template spectra , a set of comprehensive templates of known parameters covering a broad parameter space are thus of fundamental importance .
both libraries consisting of empirical and synthetic spectral templates have been used in the literature .
lists of the currently available empirical and synthetic libraries can be found in wu et al .
( 2011 ) and at a website on stellar spectral libraries .
an advantage of the empirical libraries is that they consist of spectra of @xmath11 stars .
the disadvantage is that it is often laborious and time - consuming to build an empirical spectral library of stars of accurately known parameters that cover wide parameter ranges with sufficient resolution and homogeneity .
it is clear that the parameter space encompassed by the existent empirical libraries are limited by our current knowledge of stars in the solar neighborhood and the available observations .
on the other hand , while it may be straightforward to construct a set of synthetic spectra covering homogeneously a wide parameter space , it is difficult to assess the robustness of the spectra , especially those of very low ( @xmath12k ) or high temperatures ( for instance , stars of the o , b or a spectral types ) .
properties of stellar spectra are determined not only by basic stellar parameters such as @xmath1 , log@xmath2 , [ fe / h ] and [ @xmath8/fe ] , but also depend on other parameters and processes such as the micro - turbulence and rotation velocities as well as convection .
observational uncertainties combined with inadequacies in our understanding of stellar atmospheres may lead to unrealistic parameters @xmath1 , log@xmath2 , [ fe / h ] and [ @xmath8/fe ] by matching an observed medium - to - low resolution spectrum with a library of synthetic spectra .
thanks to the efforts involving many observers , several empirical spectral libraries , including the elodie ( prugniel & soubiran 2001 ; prugniel et al .
2007 ) and miles ( snchez - blzquez et al . 2006 ; falcn - barroso et al . 2011 ) , are now available .
they cover a wide range of stellar parameters , accurately determined with high resolution spectroscopy .
a pipeline , the lamost stellar parameter pipeline ( lasp ) , which is mainly based on the universit de lyon spectroscopic analysis software ( ulyss ; koleva et al . 2009 ; wu et al .
2011 ) and makes use of the elodie library , has been developed and applied to the lamost spectra at the lamost operation and development center of the national astronomical observatories of chinese academy of sciences ( naoc ; wu et al . 2014 ) .
the library is used to determine radial velocities as well as atmospheric parameters .
parameters thus determined have been made available via the lamost official data release ( luo et al .
2012 , bai et al .
2014 ) . as parts of the lss - gac survey , a pipeline ,
the lamost stellar parameter pipeline at peking university
lsp3 , has been developed in parallel .
similar to the lasp , the lsp3 determines stellar atmospheric parameters by template matching , but using the miles rather than the elodie empirical library instead .
compared to the elodie spectra which are secured using an echelle spectrograph with a very high spectral resolution ( @xmath13 ; prugniel & soubiran 2001 ; prugniel et al .
2007 ) , the miles spectra are obtained using a long - slit spectrograph at a spectral resolution ( fwhm@xmath14 ) comparable to that of the lamost spectra , and are accurately flux - calibrated to an accuracy of a few per cent over the @xmath1535257410 wavelength coverage .
the stellar atmospheric parameters of miles spectra , determined in most cases using high resolution spectroscopy , have been calibrated to a uniform reference ( cenarro et al .
2007 ) . on the other hand ,
the radial velocities of miles stars are not as accurately determined as those in the elodie library , given the fairly low spectral resolution of miles spectra .
thus for radial velocity determinations , the lsp3 continues to make use of the elodie library .
the lsp3 has been successfully applied to hundreds of thousands spectra collected for the lss - gac survey .
radial velocities and atmospheric parameters , together with other additional parameters such as estimates of interstellar extinction and distance to individual stars are released as value - added catalogs supplementary to the lamost official data release ( paperiii ) . in this work ,
we introduce the algorithm and implementation of lsp3 in detail , and examine the accuracy of stellar parameters yielded by the lsp3 , by applying the lsp3 to the spectral templates themselves , to lamost multi - epoch spectra of duplicate stars and to lamost and sdss spectra of member candidates of open and globular clusters .
parameters yielded by the lsp3 are compared extensively with independent determinations from a number of external databases , including the pastel archive and the apogee , sdss and rave surveys , as well as with values published in the lamost first data releases ( dr1 ; bai et al .
2014 ) .
the paper is organized as follows . in section2
, we introduce template libraries adopted by the lsp3 .
section3 describes the methodology of lsp3 in detail . in section4
, we examine the lsp3 algorithm by applying it to the template spectra themselves . in section5
, we discuss the precision and accuracy of lsp3 by comparing the results yielded by different algorithms and by multi - epoch observations of duplicate stars . in section6 , lsp3 stellar parameters are compared extensively with independent determinations from external databases .
calibration and error estimates of lsp3 parameters are presented in section7 . in section8
, we discuss the error sources of lsp3 stellar parameters .
we close with a summary in section9 .
the miles library consists of 985 stars spanning wide range of stellar atmospheric parameters .
the spectra are obtained with the 2.5 m issac newton telescope , covering the wavelength range 3525 7500 at an almost constant resolution of full width at half maximum ( fwhm ) of about 2.5 ( snchez - blzquez et al .
2006 ; falcn - barroso et al . 2011 ) , which is slightly smaller than the typical fwhm of lamost spectra ( @xmath62.8 ) .
the high accuracy of ( relative ) flux calibration ( snchez - blzquez et al . 2006 ) and wide coverage of stellar parameters that are homogeneously calibrated ( cenarro et al . 2007 ) make the miles an ideal empirical spectral library for stellar parameter determinations .
the spectra are converted to match the lamost resolution by convolving with gaussians and resampled to 1.0 per pixel .
the width of the gaussian , which is allowed to vary with wavelength , is taken to be the mean of all the 4000 fibers of lamost .
the miles spectra are wavelength - calibrated to an accuracy of only approximately 10kms@xmath3 , not good enough for the purpose of radial velocity determinations for the lamost spectra .
we have thus decided to use the elodie library as radial velocity templates .
the library contains 1959 high - resolution spectra of 1388 stars , obtained with the elodie echelle spectrograph mounted on the observatoire de haute - provence 1.93 m telescope , covering wavelength range 3900 6800 at a resolving power of 42,000 ( prugniel & soubiran , 2001 ; prugniel et al .
in addition to spectra of the original resolving power , the library also provides another set of spectra , degraded to a resolving power of 10,000 .
we use the latter set of spectra .
the spectra are further degraded in resolution to match that of the lamost and resampled to 1.0 per pixel . for stars with multiple spectra ,
only the one flagged as the best is used .
note that both the miles and elodie libraries provide spectra in rest laboratory wavelengths , calculated using radial velocities determined from the spectra . as a test of the accuracy of radial velocities adopted by the elodie , we cross
- correlate the spectra with synthetic ones ( munari et al .
2005 ) of identical stellar atmospheric parameters , and find an average velocity residual and standard deviation of @xmath160.7kms@xmath3 .
the small value of standard deviation reflects the high resolution of elodie spectra and that the spectra are wavelength - calibrated to high a precision .
the offset , @xmath17kms@xmath3 , is however significant .
its origin is unclear . as shall be shown in section6.1 , we correct for any systematics in radial velocities determined with the elodie by calibrating the results against external databases . for comparison , a similar exercise for the miles library
yields a residual of @xmath186.5kms@xmath3 .
the above exercise also finds a few spectra in the elodie library that have very large velocity residuals .
for spectra with residuals in excess of 3@xmath19 of the mean , we have applied corrections to the wavelengths using the above determined residuals .
finally , given the scarce of stars of temperatures higher than 7000k , we have added 360 synthetic spectra ( munari et al .
2005 ) with temperatures between 7000 and 12,000k to the elodie library as radial velocity templates .
although the miles library has a decent coverage of the stellar parameter space , the coverage is not homogeneous and there are clusters and holes in the distribution of stars in the parameter space .
fig.1 shows the distributions of miles stars in the @xmath1 log@xmath2 and @xmath1 [ fe / h ] planes . at @xmath205700k , for example , a number of stars cluster around [ fe / h ] @xmath60.1 and @xmath6@xmath21dex , but few at [ fe / h ] @xmath6@xmath22dex .
the presence of clusters and holes in the distributions introduces patterns and biases in the resultant stellar atmospheric parameters derived by template matching .
an observational campaign to fill the holes and to further expand the parameter space coverage , as well as to expand the template spectral wavelength coverage to 9200 to utilize the full potential of the lamost spectra , especially those from the red - arm , is well under way , using the naoc 2.16 m telescope and the 2.4 m telescope of the yunnan astronomical observatory . as a remedy for the time being ,
we interpolate the miles spectra to fill up the apparent holes in parameter space .
to do this , we first exclude 85 out of the 985 miles template stars that do not have a complete set of high quality parameters ( @xmath1 , log@xmath2 and [ fe / h ] ) . of the 900 stars left ,
14 fall close the low log@xmath2 boundaries of the distribution in the @xmath1
log@xmath2 plane and are also not used for the interpolation . to interpolate the spectra ,
the remaining 886 stars are divided into four groups in the @xmath1 log@xmath2 plane ( table1 ) . for each group of stars ,
a third - order polynomial of 20 coefficients , is used to fit the spectral flux density normalized to unity at 5400 at each wavelength as a function of stellar atmospheric parameters , @xmath1 , log@xmath2 and [ fe / h ] . here
a third - order polynomial is selected as a compromise considering the fact that stellar spectra are a complicated function of atmospheric parameters and the limited number as well as parameter coverage of the miles templates . to examine the uncertainties of interpolated spectra
, we have applied the interpolation scheme to the templates themselves .
specifically , we drop a template from the library , and fit the rest with the polynomial .
the fit is then used to calculate the dropped spectrum at the given parameters . to characterize the goodness of fit
, the dispersion of the relative differences between the interpolated and observed spectra , is calculated between 4320 5500 , the window the lsp3 adopts for estimating stellar parameters . when calculating the relative differences , the interpolated spectrum is allowed to scale with a third - order polynomial to match the sed of the observed one .
the exercise is repeated for all template in the library .
for fgk dwarfs , giants , cool dwarfs and hot stars , the mean and scatter of dispersion thus calculated for all templates in the library are [email protected] , [email protected] , [email protected] and [email protected] , respectively .
fig.2 compares the interpolated and observed template spectra for four example stars , one from each of the four groups .
compared to typical uncertainties of lamost spectra , the errors of interpolated templates due to fitting uncertainties are marginal , except for lamost spectra of very high snrs . we have also tried to interpolate the templates using the formula of prugniel et al .
the results are generally comparable with each other . to fill up some of the apparent holes in parameter space covered by the miles library , some fiducial spectra are created by interpolating the existing templates in parameter space . in total
, 416 fiducial spectra are interpolated using the fits generated above and added to the miles library .
the distributions of parameters of the interpolated spectra are over - plotted in fig.1 along with those of the original spectra . the resultant parameter coverage in @xmath1 [ fe / h ] plane , though still not fully homogeneous , is much improved . finally , for the purpose of spectral classification only
, we have added 18 spectra of white dwarfs ( wds ) , carbon stars and late - m / l type stars retrieved from the sdss database to the final library of spectral templates used by the lsp3 . | we introduce the lamost stellar parameter pipeline at peking university lsp3 , developed and implemented for the determinations of radial velocity @xmath0 and stellar atmospheric parameters ( effective temperature @xmath1 , surface gravity log@xmath2 , metallicity [ fe / h ] ) for the lamost spectroscopic survey of the galactic anti - center ( lss - gac ) .
we describe the algorithms of lsp3 and examine the accuracy of parameters yielded by it .
the precision and accuracy of parameters yielded are investigated by comparing results of multi - epoch observations and of candidate members of open and globular clusters , with photometric calibration , as well as with independent determinations available from a number of external databases , including the pastel archive , the apogee , sdss and rave surveys , as well as those released in the lamost dr1 .
the uncertainties of lsp3 parameters are characterized and quantified as a function of the spectral signal - to - noise ratio ( snr ) and stellar atmospheric parameters .
we conclude that the current implementation of lsp3 has achieved an accuracy of 5.0kms@xmath3 , 150k , 0.25dex , 0.15dex for the radial velocity , effective temperature , surface gravity and metallicity , respectively , for lss - gac spectra of fgk stars of snrs per pixel higher than 10 .
the lsp3 has been applied to over a million lss - gac spectra collected hitherto .
stellar parameters yielded by the lsp3 will be released to the general public following the data policy of lamost , together with estimates of the interstellar extinction @xmath4 and stellar distances , deduced by combining spectroscopic and multi - band photometric measurements using a variety of techniques .
[ firstpage ] galaxy : disk stars : abundance techniques : spectroscopy techniques : radial velocity surveys |
let @xmath3 be a fano manifold , and let @xmath4 be a dominating family of rational curves on @xmath3 .
by this we mean that @xmath4 is an irreducible component of @xmath5 , the scheme parametrizing integral rational curves on @xmath3 , and that the union of the curves parametrized by @xmath4 is dense in @xmath3 .
we say that @xmath4 is _ locally unsplit _ if for general @xmath6 , the subfamily @xmath7 of curves containing @xmath8 is proper .
this is true , for instance , if @xmath4 is a dominating family with minimal degree with respect to some ample line bundle on @xmath3 . when @xmath4 is locally unsplit , the anticanonical degree of the curves of the family can vary between @xmath9 and @xmath10 , where @xmath1 is the dimension of @xmath3 .
following miyaoka @xcite , we define @xmath11 to be the minimal anticanonical degree of a locally unsplit dominating family of rational curves in @xmath3 , so that @xmath12 .
equivalently , @xmath11 is the minimal anticanonical degree of a free rational curve in @xmath3 ( see remark [ deflx ] ) . in this paper
we study fano manifolds with a locally unsplit dominating family of rational curves of anticanonical degree @xmath1 , including in particular fano manifolds @xmath3 with @xmath13 .
let us recall the following results , due respectively to cho , miyaoka , and shepherd - barron ( see also @xcite ) , and to miyaoka .
[ cmsbk ] let @xmath3 be a fano manifold of dimension @xmath1 .
the following properties are equivalent : 1 .
@xmath3 has a locally unsplit dominating family of rational curves of maximal anticanonical degree @xmath10 ; 2 .
@xmath14 . in particular , @xmath15 if and only if @xmath14 .
[ m ] let @xmath3 be a fano manifold of dimension @xmath16 , and with picard number @xmath17
. then @xmath13 if and only if @xmath3 is isomorphic to a quadric . on the other hand , there are also cases where @xmath13 and @xmath18 .
[ examplem ] let @xmath19 be a smooth subvariety , of dimension @xmath20 and degree @xmath21 , contained in a hyperplane .
let @xmath3 be the blow - up of @xmath22 along @xmath23 .
then @xmath3 is fano with @xmath24 and @xmath25 .
the locally unsplit dominating family of rational curves of anticanonical degree @xmath1 is given by the transforms of lines intersecting @xmath23 in one point .
first of all , we show that in fact these are the only examples .
[ result ] let @xmath3 be a fano manifold of dimension @xmath16 , with @xmath18 and @xmath13
. then there exists a smooth subvariety @xmath19 of dimension @xmath20 and degree @xmath21 , contained in a hyperplane , such that @xmath3 is isomorphic to the blow - up of @xmath22 along @xmath23 .
together with miyaoka s result ( theorem [ m ] ) , this gives a complete classification of fano manifolds with @xmath13 .
then we turn to the case where @xmath3 is a fano manifold having a locally unsplit dominating family of rational curves of anticanonical degree @xmath1 , where @xmath16 is the dimension of @xmath3 .
notice that this assumption is easier to check than the condition @xmath13 , as it involves only one family of rational curves . in the toric case ,
these varieties have been classified by fu and hwang .
let @xmath3 be a toric fano manifold , of dimension @xmath16 , having a locally unsplit dominating family of rational curves of anticanonical degree @xmath1 .
then @xmath3 is one of the following : 1 .
the blow - up of @xmath22 at a linear @xmath26 ( here @xmath27 and @xmath13 ) ; 2 .
@xmath28 ( here @xmath27 and @xmath29 ) ; 3 .
the blow - up of @xmath22 at @xmath30 , where @xmath23 is a linear @xmath26 , and @xmath31 a point not in @xmath23 ( here @xmath32 and @xmath29 ) .
we show that if @xmath3 has a locally unsplit dominating family @xmath4 of rational curves of anticanonical degree @xmath1 , then the picard number of @xmath3 is at most @xmath2 ( see proposition [ prop_main ] ) .
moreover we classify all cases with @xmath33 , giving a complete description of @xmath3 and @xmath4 .
let us describe our results .
we first contruct and study a family of examples . [ ex ] fix integers @xmath34 and @xmath35 such that @xmath16 , @xmath36 and @xmath37 .
let moreover @xmath38 be a smooth hypersurface of degree @xmath35 .
set @xmath39 , and let @xmath40 be a section of the @xmath41-bundle @xmath42 with normal bundle @xmath43
. finally set @xmath44 ( so that @xmath45 ) , and let @xmath46 be the blow - up of @xmath47 .
then @xmath3 is smooth of dimension @xmath1 and picard number @xmath2 , and it is fano if and only if @xmath48 and @xmath49 . in the fano case
, these varieties appear in @xcite , where the author classifies fano manifolds containing a divisor @xmath50 and with negative normal bundle .
[ examples ] let @xmath3 be as in example [ ex ] .
then @xmath3 has a locally unsplit dominating family @xmath4 of rational curves of anticanonical degree @xmath1 .
then we show that these are all the examples with @xmath51 .
[ main ] let @xmath3 be a fano manifold of dimension @xmath16 , and suppose that @xmath3 has a locally unsplit dominating family @xmath4 of rational curves of anticanonical degree @xmath1
. then @xmath52 . if moreover @xmath33 , then @xmath3 is isomorphic to one of the varieties described in example [ ex ] , and the family @xmath4 is unique .
we study in more detail the family of curves given by proposition [ examples ] . for @xmath6 , we denote by @xmath7 the normalization of the closed subset of @xmath4 parametrizing curves containing @xmath8 .
[ v_x ] let @xmath3 and @xmath4 be as in proposition [ examples ] , and @xmath6 a general point .
let @xmath53 be the image of @xmath8 under the morphism @xmath54 , and let @xmath55 be the degree @xmath35 morphism induced by the linear projection @xmath56 from @xmath57 .
then @xmath7 is smooth and connected . if @xmath58 , then @xmath59 . if @xmath60 , then @xmath7 is isomorphic to the relative hilbert scheme @xmath61}(a/\mathbb{p}^{n-2})$ ] of zero - dimensional subschemes , of length @xmath62 , of fibers of @xmath63 .
finally , we consider the variety of minimal rational tangents ( vmrt ) associated to the locally unsplit family @xmath4 at a general point @xmath8 , defined as follows .
let @xmath64 be the map sending a curve from @xmath7 to its tangent direction at @xmath8 , and define the vmrt @xmath65 to be the image of @xmath66 in @xmath67 .
we still denote by @xmath66 the induced map @xmath68 ; this is in fact the normalization morphism by @xcite and @xcite .
[ tau_x ] let @xmath3 and @xmath4 be as in proposition [ examples ] , and @xmath6 a general point . 1
. the vrmt @xmath69 is an irreducible hypersurface of degree @xmath70 .
if @xmath71 , then @xmath72 is an isomorphism .
3 . if @xmath73 , then @xmath72 is not an isomorphism . more precisely : the closed subset where @xmath66 is not an isomorphism has codimension @xmath74 , and the closed subset where @xmath66 is not an immersion has codimension @xmath9 .
this provides the first examples of locally unsplit dominating families of rational curves whose vmrt at a general point @xmath8 is singular , equivalently such that @xmath75 is not an isomorphism ( see ( * ? ? ?
* question 1 ) , ( * ? ? ?
* problem 2.20 ) , and @xcite ) .
notice that if @xmath76 , then @xmath4 is not a dominating family of rational curves of minimal degree ( see remark [ minimaldegree ] ) .
in order to prove theorems [ v_x ] and [ tau_x ] , we are led to study relative hilbert schemes of zero - dimensional subschemes of the projection of a smooth hypersurface from a general point .
we show the following result , of independent interest .
[ app ] fix integers @xmath77 , @xmath62 , and @xmath35 , such that @xmath78 and @xmath79 .
let @xmath80 be a smooth hypersurface of degree @xmath35 .
let @xmath81 be a general point , and let @xmath82 be the morphism induced by the linear projection from @xmath57 , where we see @xmath83 as the variety of lines through @xmath57 in @xmath84 . 1 .
the scheme @xmath61}(a/\mathbb{p}^{m})$ ] is connected and smooth of dimension @xmath77 , and the natural morphism @xmath85}(a/\mathbb{p}^{m})\to \mathbb{p}^{m}$ ] is finite of degree @xmath70 .
2 . let @xmath86 be a line through @xmath57 , and let @xmath87\in \textup{hilb}^{[a]}(a/\mathbb{p}^{m})$ ] be a point over @xmath88\in{\mathbb{p}}^m$ ] .
then @xmath89 is smooth at @xmath87 $ ] if and only if @xmath90 is a union of irreducible components of @xmath91 .
this result can be applied to study the ramification of the projection of the hypersurface @xmath23 from a general point .
the possible non - reduced fibers occurring in such a projection are well - known when @xmath23 is a curve or a surface ( see for instance @xcite ) , but not in general , to our knowledge .
we obtain the following result , which nevertheless is not optimal ( see the discussion on the case @xmath92 in the last part of the proof of theorem [ tau_x ] ) .
[ multiplicity ] let @xmath80 be a smooth hypersurface , and let @xmath82 be the morphism induced by the linear projection from a general point of @xmath84 .
let @xmath93 be any fiber
. then @xmath94\le m.\ ] ] * outline of the paper . * in section [ notations ] , we introduce notations used in the remainder of the paper , and we discuss some properties of families of rational curves .
in section [ d ] we study fano manifolds @xmath3 of dimension @xmath16 having a prime divisor @xmath95 with @xmath96 , using techniques from the minimal model program , and in particular results from @xcite .
it follows from ( * ? ? ? * proposition 5 ) that @xmath52 ; we study the cases @xmath24 and @xmath33 . in the case
@xmath24 , we describe the possible extremal contractions of @xmath3 ( see remark [ nonnefdiv ] and proposition [ divisorial ] ) . then we give a complete classification of these varieties when @xmath33 , see example [ ex2 ] and theorem [ main2 ] .
this generalizes results from @xcite for the case @xmath97 and negative normal bundle . in section [ proofs ]
, we specialize the results of the previous section to the case where @xmath3 has a locally unsplit dominating family @xmath4 of rational curves of anticanonical degree @xmath1 .
indeed , for general @xmath6 , any irreducible component of the locus swept out by curves of the family through @xmath8 is a divisor @xmath95 with @xmath96 .
this yields @xmath52 .
we first prove theorem [ result ] on the case @xmath13 .
next we show that , under the assumptions of theorem [ main ] , if moreover @xmath33 , then @xmath3 is isomorphic to one of the varieties described in example [ ex ] , and we determine the class of the family @xmath4 in @xmath98 ( see proposition [ prop_main ] ) . finally , when @xmath24 , we describe the possible extremal contractions of @xmath3 ( see lemma [ contraction ] ) . in this section
we use repeatedly the characterization of the projective space given by theorem [ cmsbk ] . in section [ families ]
we first construct a locally unsplit dominating family of rational curves on the varieties introduced in example [ ex ] , proving proposition [ examples ] . with the use of proposition [ prop_main ]
, we also show that the family is unique , completing the proof of theorem [ main ] . finally , we show theorems [ v_x ] and [ tau_x ] , using theorem [ genus_hilbert ] from the appendix ( section [ appendix ] ) . in section [ appendix ]
we discuss the relative hilbert scheme @xmath61}(a/\mathbb{p}^{m})$ ] , where @xmath99 is the projection of a smooth hypersurface @xmath100 from a general point .
we first study local properties of the natural morphism @xmath101}(a/\mathbb{p}^{m})\to{\mathbb{p}}^m$ ] , and we show that @xmath61}(a/\mathbb{p}^{m})$ ] is integral ( see theorem [ genus_hilbert ] ) .
this is used in the proof of theorem [ v_x ] . in theorem [ genus_hilbert ]
we also determine the genus of the curve @xmath61}(a/\mathbb{p}^{1})$ ] ; together with the description of the fibers of @xmath89 , this is crucial for the proof of theorem [ tau_x ] .
at last we prove that the hilbert scheme is smooth ( theorem [ app ] ) , as a consequence of theorem [ v_x ] , and we show corollary [ multiplicity ]
. * acknowledgements .
* this work was initiated during the conference _ gomtrie algbrique et gomtrie complexe _ at c.i.r.m . ,
marseille , in march 2012 ; we thank the organizers for this opportunity .
we are also grateful to alberto collino for some useful conversations .
the authors were partially supported by the european research network program gdre - grifga ( _ italian - french gdre in algebraic geometry _ ) .
the first named author was partially supported by the research project m.i.u.r .
prin 2009 `` spazi di moduli e teoria di lie '' .
the second named author was partially supported by the project _ class _ of agence nationale de la recherche , under agreement anr-10-jcjc-0111 .
we would like to thank these institutions for their support .
throughout this paper , we work over the field of complex numbers .
we will use the definitions and apply the techniques of the minimal model program frequently , without explicit references .
we refer the reader to @xcite and @xcite for background and details . for any projective variety @xmath3 ,
we denote by @xmath98 ( respectively , @xmath102 ) the vector space of one - cycles ( respectively , cartier divisors ) , with real coefficients , modulo numerical equivalence .
we denote by @xmath103 $ ] ( respectively , @xmath104 $ ] ) the numerical equivalence class of a curve @xmath105 ( respectively , of a cartier divisor @xmath95 ) .
moreover , @xmath106 is the convex cone generated by classes of effective curves .
_ for any closed subset @xmath107 , we denote by @xmath108 the subspace of @xmath98 generated by classes of curves contained in @xmath109 . _
if @xmath95 is a cartier divisor in @xmath3 , we set @xmath110 .
if @xmath3 is a normal projective variety , a _ contraction _ of @xmath3 is a surjective morphism @xmath111 , with connected fibers , where @xmath112 is normal and projective .
the contraction is elementary if @xmath113 .
let @xmath114 be an extremal ray of @xmath115 .
if @xmath95 is a divisor in @xmath3 , the sign of @xmath116 is the sign of @xmath117 , @xmath118 a non - zero one - cycle with class in @xmath114 .
suppose that @xmath119 , and let @xmath120 be the associated elementary contraction .
we set @xmath121 , the locus where @xmath122 is not an isomorphism . by a @xmath41-bundle we mean a smooth morphism whose fibers are isomorphic to @xmath41 , while a morphism is called a conic bundle if every fiber is isomorphic to a plane conic . if @xmath123 is a vector bundle on a variety @xmath112 , we denote by @xmath124 the scheme @xmath125 .
we refer the reader to ( * ? ? ? * and iv.2 ) for the main properties of families of rational curves ; we will keep the same notation as @xcite .
in particular , we recall that @xmath5 is the normalization of the open subset of @xmath126 parametrizing integral rational curves . by a family of rational curves in @xmath3 we mean an irreducible component @xmath4 of @xmath5 .
we say that @xmath4 is a _ dominating family _ if its universal family dominates @xmath3 .
we say that @xmath4 is _ locally unsplit _ if , for a general point @xmath6 , the subfamily of @xmath4 parametrizing curves through @xmath8 is proper .
let @xmath4 be a locally unsplit dominating family of rational curves on @xmath3 .
the class in @xmath98 of a curve @xmath105 from @xmath4 does not depend on the choice of @xmath105 , and will be denoted by @xmath127 $ ] .
we denote by @xmath103\in v$ ] a point corresponding to the integral rational curve @xmath128 .
for @xmath6 , we denote by @xmath7 the _ normalization _ of the closed subset of @xmath4 parametrizing curves through the point @xmath8 , and by @xmath129 the union of all curves of the family @xmath7 .
we denote by @xmath130 the scheme parametrizing morphisms from @xmath41 to @xmath3 sending @xmath131 to @xmath8 .
we now recall some well - known properties .
suppose that @xmath6 is _
general_. then every curve in @xmath7 is free .
this implies that @xmath7 is smooth , of dimension @xmath132 - 2 $ ] , but possibly not connected .
moreover there is a smooth closed subset @xmath133 , which is a union of irreducible components , containing all birational maps @xmath134 such that @xmath135 and @xmath136 is a curve of the family @xmath7 .
there is an induced smooth morphism @xmath137 , sending @xmath138 $ ] to @xmath139 $ ] . still for a general point @xmath8
, if a curve from @xmath7 is smooth at @xmath8 , then it is parametrized by a unique point of @xmath7 .
we say that an integral rational curve @xmath128 is _ standard _ if the pull - back of @xmath140 under the normalization @xmath141 is @xmath142 for some @xmath143 .
in this section we study fano manifolds @xmath3 having a prime divisor @xmath95 with @xmath96 , or more generally @xmath144 .
the main technique here is the study of extremal rays and contractions of @xmath3 .
the first step for the proofs of theorems [ result ] and [ main2 ] is the following lemma .
it is a standard application of mori theory , in particular the proof can be adapted from ( * ? ? ?
* proof of proposition 5 ) , and follows the same strategy used in @xcite .
we give a short proof for the reader s convenience .
[ uno ] let @xmath3 be a fano manifold of dimension @xmath16 and picard number @xmath18 , and let @xmath145 be a prime divisor with @xmath144 .
then one of the following holds : 1 .
@xmath24 and there exists a blow - up @xmath46 with center @xmath146 smooth of codimension @xmath9 , @xmath112 is smooth and fano , and @xmath147 , where @xmath148 is the contraction of the extremal ray @xmath114 of @xmath115 ; 2 . @xmath24 and there exists a conic bundle @xmath46 , finite on @xmath95 , such that @xmath112 is smooth and fano ; 3 .
@xmath33 and there is a conic bundle @xmath149 , finite on @xmath95 , such that @xmath109 is smooth , fano , and @xmath150 . the conic bundle @xmath122 is the contraction of a face @xmath151 of @xmath115 , where @xmath114 and @xmath152 both correspond to a smooth blow - up of a codimension @xmath9 subvariety .
moreover @xmath147 , and we have a diagram : @xmath153_{{\widehat}{\sigma}}\ar[dd]^{{\varphi}}\ar[dr]^{\sigma}&\\ { { \widehat}{y}}\ar[dr]_{{\widehat}{\pi}}&&y\ar[dl]^{\pi}\\ & z & } \ ] ] where @xmath148 is the contraction of @xmath114 , @xmath154 is the contraction of @xmath152 , @xmath112 and @xmath155 are smooth , @xmath112 is fano , the center @xmath146 of the blow - up @xmath148 is contained in @xmath156 , and @xmath157 and @xmath158 are conic bundles .
let @xmath114 be an extremal ray of @xmath115 with @xmath147 , and @xmath46 the associated contraction .
if @xmath159 , then every curve contained in @xmath95 has class in @xmath114 , hence @xmath160 is a point and @xmath161 . we conclude that @xmath162 , because @xmath147 . on the other hand , since @xmath18 , we can find a non - trivial fiber @xmath163 of @xmath148 disjoint from @xmath95 , which yields @xmath164 , a contradiction .
therefore @xmath165 .
this implies that @xmath148 is finite on @xmath95 , therefore every non - trivial fiber of @xmath148 has dimension one .
thus @xmath112 is smooth and there are two possibilities : either @xmath148 is a conic bundle , or it is the blow - up of @xmath146 with @xmath47 smooth of codimension @xmath9 ( see ( * ? ? ? * theorem 1.2 ) ) . if @xmath148 is a conic bundle , then @xmath166 because @xmath167 , so we are in case @xmath168 . if @xmath148 is a blow - up and @xmath166 , then we are in case @xmath169 .
assume that @xmath148 is a blow - up and @xmath170 , and set @xmath156 .
then @xmath171 is a prime divisor in @xmath112 and @xmath172 , hence @xmath173 .
moreover @xmath174 .
let @xmath175 be the exceptional divisor .
we have @xmath176 .
if @xmath177 is an irreducible curve , and @xmath178 is an irreducible curve not contained in @xmath47 , then there exists @xmath179 such that @xmath180 , so that @xmath181 , where @xmath182 is the transform of @xmath183 in @xmath3 .
this implies that @xmath112 is fano ( see @xcite ) .
we repeat the same argument in @xmath112 and take an extremal ray @xmath184 with @xmath185 .
similarly as before we see that @xmath186 , so that again , if @xmath187 is the contraction of @xmath188 , @xmath157 is finite on @xmath171 and has fibers of dimension at most one .
hence as before @xmath157 is either a conic bundle or the smooth blow - up of a subvariety of codimension @xmath9 in @xmath109 . if @xmath157 is a conic bundle , then @xmath150 because @xmath189 , so @xmath33 .
set @xmath190 ; then @xmath122 has a second factorization @xmath191 . since every fiber of @xmath122 has dimension one , both @xmath154 and @xmath158 have fibers of dimension @xmath192 . applying ( * ? ? ?
* theorem 4.1 ) we conclude that @xmath122 and @xmath158 are conic bundles , @xmath155 is smooth , and @xmath154 the blow - up of a smooth subvariety of codimension @xmath9 , so we are in case @xmath193 .
finally , @xmath157 can not be a blow - up .
indeed if so , @xmath194 is a prime divisor which intersects @xmath171 , and since @xmath173 , @xmath194 has strictly positive intersection with every curve contained in @xmath171 .
in particular @xmath194 must intersect @xmath47 , as @xmath195 .
if @xmath163 is a non - trivial fiber of @xmath157 with @xmath196 , and @xmath197 is its transform , one has @xmath198 , a contradiction .
remark [ nonnefdiv ] and proposition [ divisorial ] below describe the possible extremal contractions of @xmath3 in the case @xmath24 .
[ nonnefdiv ] let @xmath3 be a fano manifold , and @xmath145 a prime divisor with @xmath144 .
if @xmath95 is not nef , then it is the exceptional locus of a divisorial contraction sending @xmath95 to a point .
let @xmath114 be an extremal ray of @xmath115 such that @xmath199 , and @xmath148 the associated contraction .
notice that @xmath200 since @xmath199 . on the other hand , every curve contained in @xmath95 has class in @xmath114 since @xmath144 .
this implies that @xmath201 , and that @xmath160 is a point .
[ divisorial ] let @xmath3 be a fano manifold of dimension @xmath16 and picard number @xmath24 , and let @xmath145 be a nef prime divisor with @xmath144 .
then @xmath202 is an extremal ray of @xmath3 , and one of the following holds : 1 .
the contraction of @xmath203 is @xmath204 , and @xmath95 is a fiber ; 2 .
the contraction of @xmath203 is divisorial , sends its exceptional divisor @xmath205 to a point , and @xmath206 ; 3 .
the contraction of @xmath203 is small , it has a flip @xmath207 , @xmath208 is smooth , and there is a @xmath41-bundle @xmath209 .
moreover @xmath210 is finite on the transform of @xmath95 in @xmath208 .
furthermore , if there exists a smooth , irreducible subvariety @xmath211 , of codimension @xmath9 , such that the blow - up of @xmath3 along @xmath23 is fano , then @xmath193 can not happen .
_ we first notice that @xmath95 is not ample , because @xmath212 . indeed
the push - forward of one - cycles @xmath213 is not surjective , so that the restriction map @xmath214 is not injective .
we have @xmath215 ( because @xmath3 is fano ) and @xmath216 , hence the restriction map @xmath217 is not injective as well . by lefschetz theorem on hyperplane sections , @xmath95 can not be ample ( recall that @xmath16 ) . since @xmath95 is nef and non - ample , and @xmath24 ,
we conclude that @xmath218 is an extremal ray @xmath203 of @xmath115 .
set @xmath219 .
if @xmath220 , then the contraction of @xmath203 sends @xmath95 to a point , and hence @xmath221 . on the other hand @xmath222 , thus @xmath95 is the pull - back of a cartier divisor .
therefore the target of the contraction of @xmath203 is @xmath41 , @xmath95 is a fiber , and we are in case @xmath169 .
we assume that @xmath223 . if @xmath224 , then @xmath95 must intersect some irreducible curve @xmath105 with class in @xmath203 , and this yields @xmath225 , because @xmath226 .
this contradicts @xmath223 , therefore @xmath206 . in particular , the contraction of @xmath203 is birational . finally @xmath227 has dimension one ; this implies that the contraction of @xmath203 maps @xmath205 to points .
if @xmath203 is a divisorial extremal ray , then we are in case @xmath168 .
suppose now that the contraction of @xmath203 is small ; by ( * ? ? ?
* corollary 1.4.1 ) the flip @xmath207 of @xmath203 exists .
let @xmath228 be the transform of @xmath95 , @xmath229 the small extremal ray of @xmath208 associated with the flip , and @xmath230 .
notice that @xmath231 , as @xmath206 .
notice also that @xmath208 has normal , @xmath232-factorial , and terminal singularities , and @xmath233 .
we have @xmath234 and @xmath235 ; on the other hand , a curve disjoint from @xmath236 has positive anticanonical degree .
in particular , by the cone theorem , @xmath208 has a second extremal ray @xmath237 with @xmath238 , and @xmath239 since the flip @xmath207 is an isomorphism in a neighborhood of @xmath95 , the linear subspace @xmath240 stays one - dimensional .
moreover @xmath241 , hence we must have @xmath242 .
let @xmath209 be the contraction of @xmath237 .
similarly as before , we see that @xmath243 .
since the contraction of @xmath203 sends @xmath205 to points , the contraction of @xmath229 sends @xmath236 to points .
this implies that every curve in @xmath236 has class in the extremal ray @xmath229 .
we deduce that @xmath210 is finite on both @xmath244 and @xmath236 .
in particular , since @xmath242 , every non - trivial fiber of @xmath210 has dimension one . [ degrees ]
let @xmath105 be an irreducible component of a non - trivial fiber of @xmath210 . if @xmath245 , then : @xmath246 indeed if @xmath247 is the transform of @xmath105 , we have @xmath248 .
this follows from ( * ? ? ?
* lemma 3.38 ) , see ( * ? ? ?
* lemma 3.8 ) for an explicit proof .
we show that @xmath210 is of fiber type . by contradiction , assume that @xmath210 is birational .
suppose that @xmath249 , and let @xmath250 be an irreducible component of a fiber of @xmath210 which intersects @xmath236 .
notice that @xmath251 ( since @xmath210 is finite on @xmath236 ) , in particular @xmath252 .
we get @xmath253 by ( * ? ? ?
* lemma 1.1 ) , and @xmath254 by [ degrees ] , a contradiction .
therefore @xmath255 , so that @xmath256 is contained in the smooth locus of @xmath208 . by (
* ? ? ? * theorem 1.2 ) , @xmath256 is a divisor .
we have @xmath257 and @xmath258 , hence @xmath259 is nef , a contradiction .
thus @xmath210 is of fiber type .
we show that @xmath260 is a @xmath261-bundle with @xmath208 and @xmath262 smooth , so that we are in case @xmath193 .
since @xmath263 can not dominate @xmath262 , the general fiber of @xmath210 is a smooth rational curve of anticanonical degree @xmath9 .
suppose that there is a fiber @xmath163 of @xmath210 such that the corresponding one - cycle is not integral and @xmath264 .
then there is an irreducible component @xmath105 of @xmath163 , such that @xmath265 . if @xmath266 , then @xmath267 is cartier in a neighbourhood of @xmath105 , and we must have @xmath268 and @xmath269 ( where we consider @xmath163 as a one - cycle ) .
thus , up to replacing @xmath105 with another irreducible component of @xmath163 , we may assume that @xmath245 , and @xmath265 ; but this contradicts [ degrees ] . by ( * ? ? ?
* theorem ii.2.8 ) , @xmath210 is smooth in a neighbourhood of @xmath270 .
thus @xmath271 , because @xmath233 .
this implies that @xmath272 , since @xmath210 is finite on @xmath236 . in particular
, @xmath262 is smooth ( see ( * ? ? ?
* theorem 4.1(2 ) ) and references therein ) , @xmath210 is a conic bundle , and either the discriminant locus @xmath273 of @xmath210 has pure codimension one or @xmath274 .
if @xmath275 , then @xmath276 ( @xmath210 is finite on @xmath236 and @xmath277 ) , a contradiction .
this proves that @xmath260 is a @xmath261-bundle with @xmath208 and @xmath262 smooth .
suppose now that we are in case @xmath193 , and that there is a smooth irreducible subvariety @xmath211 of codimension @xmath9 , such that the blow - up of @xmath3 along @xmath23 is fano .
we show that this gives a contradiction .
let @xmath278 the transform of @xmath23 , and let us consider the divisor @xmath279 in @xmath208 . since @xmath280 , we must have @xmath281 .
therefore we find a fiber @xmath282 of @xmath210 which intersects both @xmath283 and @xmath236 .
let @xmath284 be the transform of @xmath282 .
as in [ degrees ] , we see that @xmath285 , so that @xmath286 . on the other hand @xmath287 and @xmath288 ,
hence the transform of @xmath163 in the blow - up of @xmath3 along @xmath23 should have non - positive anticanonical degree , a contradiction .
we will show that any fano manifold @xmath3 with @xmath33 having a prime divisor @xmath95 with @xmath144 is isomorphic to one of the varieties described below .
we first recall the following definition for the reader s convenience .
let @xmath289 be an ample line bundle on a normal projective variety @xmath109 .
consider the @xmath41-bundle @xmath290 , with natural projection @xmath291 .
the tautological line bundle @xmath292 is semiample on @xmath112 . for @xmath293 , the linear system @xmath294 induces a birational morphism @xmath295 onto a normal projective variety , contracting the divisor @xmath296 corresponding to the projection @xmath297 to a point . following @xcite , we call @xmath298 the _ normal generalized cone over the base @xmath299_. [ ex2 ] fix integers @xmath1 , @xmath62 , and @xmath35 , such that @xmath0 , @xmath300 , and @xmath36
. let @xmath109 be a fano manifold of dimension @xmath301 , with @xmath150 .
let @xmath302 be the ample generator of @xmath303 . if @xmath77 is an integer
, then we write @xmath304 for @xmath305 .
let moreover @xmath306 be a smooth hypersurface .
set @xmath307 , and let @xmath187 be the @xmath41-bundle . if @xmath60 , then there is a birational contraction @xmath308 sending a divisor @xmath309 to a point , where @xmath298 is the normal generalized cone over @xmath310 .
we have @xmath311 , @xmath309 is a section of @xmath157 , and @xmath312 . if @xmath58 , then @xmath313 .
let @xmath309 be a fiber of @xmath314 .
we have @xmath311 , @xmath309 is a section of @xmath157 , and @xmath315 .
let now @xmath316 be a section of @xmath157 with normal bundle @xmath317 .
notice that @xmath318 if @xmath60 .
if @xmath58 , we choose @xmath319 such that @xmath318 .
set @xmath44 .
finally let @xmath46 be the blow - up of @xmath47 ( see figure [ fig1 ] ) .
then @xmath3 is a smooth projective variety of dimension @xmath1 , with @xmath33 .
let @xmath320 be the tranforms of @xmath321 respectively .
then @xmath322 , @xmath323 , and @xmath324 .
the composition @xmath190 is a conic bundle , and has a second factorization : @xmath325_{{\widehat}{\sigma}}\ar[dr]^{\sigma}\ar[dd]^{{\varphi } } & \\ { { \widehat}{y}}\ar[dr]_{{\widehat}{\pi } } & & y\ar[dl]^{\pi}\\ & z & } \ ] ] where @xmath326 .
the images @xmath327 and @xmath328 are disjoint sections of the @xmath41-bundle @xmath329 , with normal bundles @xmath330 and @xmath331 .
moreover @xmath154 is the blow - up of @xmath155 along the intersection @xmath332 .
set @xmath333 and @xmath334 , and let @xmath335 and @xmath336 be fibers of the @xmath41-bundles .
let moreover @xmath337 be an irreducible curve having minimal intersection with @xmath302 , and set @xmath338 .
finally let @xmath339 and @xmath340 be curves corresponding to @xmath341 .
we have the following relations of numerical equivalence : @xmath342 and the relevant intersections are shown in table [ table ] , where @xmath343 is the index of @xmath109 , _
i.e. _ the integer defined by @xmath344
. @xmath345 [ ne(x ) ] the cone @xmath115 is closed and polyhedral . if @xmath58 , then @xmath115 has three extremal rays , generated by the classes of @xmath163 , @xmath346 , and @xmath347 , with loci @xmath348 , @xmath349 , and @xmath350 respectively . if @xmath351 , then @xmath115 has three extremal rays , generated by the classes of @xmath163 , @xmath346 , and @xmath352 , with loci @xmath348 , @xmath349 , and @xmath205 respectively . if instead @xmath353 , then @xmath115 is non - simplicial and has @xmath354 extremal rays , generated by the classes of @xmath163 , @xmath346 , @xmath352 , and @xmath347 , with loci @xmath348 , @xmath349 , @xmath205 and @xmath350 respectively
. set @xmath355 $ ] , @xmath356 $ ] , @xmath357 $ ] , and @xmath358 $ ] .
we already know that @xmath114 and @xmath152 are extremal rays of @xmath359 , and that @xmath360 is a face . since @xmath361 is nef and non - ample , @xmath362 is nef and non - ample in @xmath112 , and the same holds for @xmath363 in @xmath3 .
it is not difficult to see that @xmath364 . in particular
, this shows that @xmath203 is an extremal ray of @xmath359 , so that there exists a nef divisor @xmath365 such that @xmath366 . if @xmath367 , then similarly as before @xmath327 is nef and non - ample in @xmath155 , so that @xmath368 is nef in @xmath3 .
the plane @xmath369 intersects @xmath359 along the face @xmath370 ; in particular , @xmath371 is an extremal ray . finally , the divisor @xmath372 is nef , it is not numerically trivial , and @xmath373 . therefore @xmath374 , and we get the statement in the case @xmath367 . if instead @xmath375 , then @xmath328 is nef , non - ample in @xmath155 , so that @xmath376 is nef , and does not intersect @xmath205 .
we get @xmath377 , which gives the statement in the case @xmath375 .
the case @xmath58 follows from the case @xmath378 , see remark [ symm ] . a straightforward consequence of lemma [ ne(x ) ]
is the following .
@xmath3 is fano if and only if @xmath379 and @xmath380 . indeed ,
since @xmath115 is closed and polyhedral , @xmath3 is fano if and only if every extremal ray of @xmath115 has positive intersection with the anticanonical divisor ( see table [ table ] ) .
[ symm ] suppose that @xmath381 .
then by choosing @xmath382 , we get a variety @xmath208 isomorphic to @xmath3 , with the roles of @xmath112 and @xmath155 interchanged .
example [ ex ] is a special case of this example , with @xmath383 , and the additional condition @xmath381 .
we are now in position to prove the main result of this section .
[ main2 ] let @xmath3 be a fano manifold of dimension @xmath16 and picard number @xmath33 , and let @xmath145 be a prime divisor with @xmath144 .
then @xmath3 is isomorphic to one of the varieties described in example [ ex2 ] .
remark that if @xmath3 is as in example [ ex2 ] , then @xmath205 is a prime divisor with @xmath384 , and hence @xmath385 . _
proof of theorem [ main2 ] . _ [ primo ] as @xmath33 , we are in case @xmath193 of lemma [ uno ] , and there is a conic bundle @xmath149 .
we keep the same notation as in lemma [ uno ] ; in particular we recall the diagram : @xmath153_{{\widehat}{\sigma}}\ar[dd]^{{\varphi}}\ar[dr]^{\sigma}&\\ { { \widehat}{y}}\ar[dr]_{{\widehat}{\pi}}&&y\ar[dl]^{\pi}\\ & z & } \ ] ] we set @xmath386 and @xmath387 .
notice that we may have @xmath388 ( if @xmath389 ) or @xmath390 ( if @xmath391 ) . in the first case @xmath46 and @xmath392
have the same properties with respect to @xmath3 and @xmath95 , so that their role is interchangeable , while in the second case the behaviour of the two blow - ups with respect to @xmath95 is different . [ int ]
we show that every prime divisor @xmath393 must intersect @xmath394 .
we first notice that @xmath395 in @xmath112 . indeed ,
if @xmath396 , then the claim is obvious .
otherwise , @xmath397 , and the claim follows from @xmath150 .
thus , if @xmath398 , then @xmath399 intersects @xmath348 . otherwise , @xmath399 intersects @xmath349 .
[ smooth ] we show that @xmath187 is a smooth morphism .
otherwise it has a non - empty discriminant divisor @xmath400 , whose inverse image @xmath401 must be disjoint from @xmath394 since @xmath3 is fano , which contradicts [ int ] .
[ dynef ] let us consider the prime divisor @xmath402 .
we have @xmath172 , hence @xmath173 .
we show that up to replacing @xmath95 with another prime divisor @xmath403 , we can assume that @xmath171 is nef .
suppose that @xmath171 is not nef .
then @xmath171 is the exceptional locus of a divisorial contraction @xmath295 sending @xmath171 to a point , by remark [ nonnefdiv ] .
the exceptional locus of the composite map @xmath404 is @xmath405 , and is contracted to a point in @xmath298 . in particular , neither @xmath95 is nef ( see for instance ( * ? ? ?
* lemma 3.39 ) ) .
let @xmath403 be the pull - back of a general prime divisor in @xmath298 .
then @xmath244 is a prime nef divisor in @xmath3 , and @xmath406 .
notice that the classes @xmath104 $ ] and @xmath407 $ ] in @xmath102 can not be proportional , therefore the planes @xmath408 and @xmath409 in @xmath98 are distinct . as @xmath410 , we get @xmath411 . by [ int ] , we have @xmath412 , hence @xmath413 . moreover , as @xmath244 is nef , we must have @xmath414 nef too .
this shows that up to replacing @xmath95 with @xmath244 , and @xmath114 with @xmath152 , we can assume that @xmath171 is nef .
[ applydivisorial ] by [ dynef ] , we can assume that @xmath171 is nef .
then proposition [ divisorial ] applies , and @xmath415 is an extremal ray @xmath416 of @xmath417 .
moreover , since @xmath174 and the blow - up of @xmath112 along @xmath47 is fano , case @xmath193 of proposition [ divisorial ] is excluded .
suppose that we are in case @xmath169 of proposition [ divisorial ] .
then @xmath313 ( see for instance ( * ? ? ?
* lemma 4.9 ) ) , and @xmath171 is a fiber of @xmath418 . since @xmath174 , @xmath3 is isomorphic to one of the varieties described in example [ ex2 ] , with @xmath58 .
this completes the proof of theorem [ main2 ] in this case .
we assume that we are in case @xmath168 of proposition [ divisorial ] , so that the extremal ray @xmath416 is divisorial , its contraction sends the divisor @xmath419 to a point , and @xmath420 .
in particular , @xmath421 . denote by @xmath422 the ample generator of @xmath423 . by lemma [ section_bundle ] below ,
@xmath424 is a section of @xmath187 , and there exists an integer @xmath60 such that @xmath425 and @xmath426 .
[ secondsectiony ] suppose that there exists a section @xmath427 of @xmath187 , disjoint from @xmath309 , and containing @xmath47 .
we claim that this implies the statement .
indeed we have @xmath428 for some @xmath429 , and since @xmath430 , restricting to @xmath309 we get @xmath431 , and restricting to @xmath319 we get @xmath432 .
thus @xmath3 is one of the varieties described in example [ ex2 ] , for @xmath60 .
let @xmath433 be the transform of @xmath309 .
we have @xmath434 since @xmath435 , and hence @xmath436 by [ int ] .
let us consider now the image @xmath437 .
then @xmath327 is a section of @xmath158 , so that @xmath438 and @xmath439
. moreover @xmath327 contains the center @xmath440 of the blow - up @xmath154 .
this also implies that @xmath155 is fano , as in the proof of lemma [ uno ] .
suppose now that there exists a section @xmath441 of @xmath442 , disjoint from @xmath327 .
then its transform in @xmath112 yields a section of @xmath187 , disjoint from @xmath309 , and containing @xmath47 , and this implies the statement by [ secondsectiony ] . in order to construct such @xmath365
, we consider the divisor @xmath443 .
notice that @xmath444 , and that the two divisors @xmath445 and @xmath327 are distinct , because @xmath206 in @xmath3 .
[ secondsection ] suppose first that @xmath445 is not nef . by remark [ nonnefdiv ]
, @xmath445 is the exceptional locus of a divisorial contraction sending @xmath445 to a point .
then , by lemma [ section_bundle ] , @xmath445 is a section of @xmath158 .
moreover we have @xmath446 for every curve @xmath447 , because @xmath448 and @xmath439 . since @xmath445 is not nef , the divisors @xmath445 and @xmath327 must be disjoint , and we can set @xmath449 .
we assume now that @xmath445 is nef .
then proposition [ divisorial ] applies , and @xmath450 is an extremal ray @xmath451 of @xmath452 .
we claim that case @xmath193 of proposition [ divisorial ] can not happen , namely that @xmath451 can not be small .
indeed this follows from proposition [ divisorial ] if @xmath453 , namely if @xmath454 .
if instead @xmath455 , then @xmath456 , hence the contraction of @xmath451 sends @xmath327 to a point , and it can not be small .
suppose that we are in case @xmath169 of proposition [ divisorial ] .
as in [ applydivisorial ] we see that @xmath457 , and @xmath445 and @xmath327 are fibers of the projection @xmath458 .
thus we can define @xmath365 to be a general fiber of the projection @xmath459 .
finally , suppose that we are in case @xmath168 of proposition [ divisorial ] , so that the contraction of @xmath451 is divisorial . by lemma [ section_bundle ] , @xmath460 is a section of @xmath158 .
so if @xmath461 , we set @xmath462 . if instead @xmath463 , then we get @xmath464 , because @xmath439 . therefore by lemma [ section_bundle ] , there is a section @xmath365 of @xmath158 disjoint from @xmath327 .
the following result is certainly well - known to experts .
we include a proof for lack of references .
[ section_bundle ] let @xmath112 and @xmath109 be smooth connected projective varieties , and let @xmath465 be a @xmath41-bundle .
let @xmath466 be a birational morphism onto a projective variety sending an effective and reduced divisor @xmath205 to points .
then @xmath467 for some line bundle @xmath468 on @xmath109 so that @xmath205 identifies with the section of @xmath157 corresponding to @xmath469 .
moreover , @xmath470 is ample . by replacing @xmath210 with its stein factorization
, we may assume that @xmath298 is normal and that @xmath210 has connected fibers .
we show that @xmath205 is a section of @xmath471 .
notice that @xmath157 is finite on @xmath205 .
let @xmath472 be a general smooth connected curve , and set @xmath473 .
then @xmath474 is a reduced curve ; let @xmath105 be an irreducible component of @xmath474
. moreover let @xmath475 be a minimal section , and @xmath476 a fiber of @xmath157 .
then @xmath477 . set @xmath478 .
suppose that @xmath479 .
then @xmath480 where @xmath481 , @xmath482 if @xmath483 , and @xmath484 if @xmath485 ( see ( * ? ? ?
* propositions v.2.20 and v.2.21 ) ) .
thus : @xmath486 on the other hand , the restriction of @xmath210 to @xmath203 induces a birational morphism @xmath487 sending @xmath105 to a point , and hence @xmath488 , yielding a contradiction .
thus @xmath489 , hence @xmath490 .
this completes the proof of the first assertion .
set @xmath491 .
then @xmath492 is a locally free sheaf of rank @xmath9 that fits into a short exact sequence @xmath493 corresponding to @xmath494 , @xmath112 identifies with @xmath495 , @xmath496 with the tautological line bundle @xmath497 , and @xmath205 corresponds to @xmath498 . by (
* lemme 3.2 and lemme 3.3 ) we have @xmath499 and @xmath500 $ ] maps to @xmath501 under the above isomorphism .
let @xmath502 be an ample line bundle on @xmath298 .
then there exists @xmath503 such that @xmath504 .
this implies that @xmath505=0\in \textup{pic}(2g)$ ] .
hence we must have @xmath506 , and @xmath507 .
let @xmath236 be the section of @xmath471 corresponding to @xmath508 .
then @xmath509 , therefore @xmath510 is ample . but @xmath511 under the isomorphism @xmath512 .
this completes the proof of the lemma .
in this section , we prove theorem [ result ] and proposition [ prop_main ] . we start with the following observations .
[ locuns ] let @xmath4 be a locally unsplit dominating family of rational curves on a smooth projective variety @xmath3 , and suppose that the curves of the family have anticanonical degree @xmath1 .
let @xmath8 be a general point , and let @xmath129 be the union of all curves parametrized by @xmath7 .
then by ( * ? ? ? * corollaries iv.2.6.3 and ii.4.21 ) we have that @xmath513 is a divisor and @xmath514 $ ] . [ deflx ]
let @xmath3 be a fano manifold , and recall that @xmath11 is the minimal anticanonical degree of a locally unsplit dominating family of rational curves in @xmath3 .
if @xmath6 is a general point , then every irreducible rational curve @xmath105 through @xmath8 has anticanonical degree at least @xmath11 , see ( * ? ? ?
* theorem iv.2.4 ) .
this implies that @xmath11 can equivalently be defined as the minimal anticanonical degree of a dominating family of rational curves in @xmath3 . since @xmath13
, there is a locally unsplit dominating family @xmath4 of rational curves of anticanonical degree @xmath1 .
thus by remark [ locuns ] @xmath3 contains a prime divisor @xmath95 with @xmath144 , so that we can apply lemma [ uno ] . since @xmath515 , we know that @xmath3 can not have a conic bundle structure
. therefore lemma [ uno ] yields that @xmath24 and there exists @xmath46 such that @xmath112 is smooth with @xmath516 and @xmath166 , and @xmath148 is the blow - up of @xmath517 smooth of codimension @xmath9 . let @xmath175 be the exceptional divisor ; we have @xmath176 .
let @xmath90 be a locally unsplit dominating family of rational curves in @xmath112 , @xmath518 a general curve of the family , and @xmath247 its transform .
then @xmath519 , hence @xmath520 .
moreover @xmath182 moves in a dominating family of rational curves in @xmath3 , so that @xmath521 .
this yields @xmath522 .
we show that @xmath523 > n$ ] . indeed if @xmath523=n$ ] , then for any curve @xmath105 of the family such that @xmath519
, we must have @xmath524 , and hence @xmath525 . on the other hand , for @xmath526 general
, @xmath527 is an ample divisor ( recall that @xmath166 ) , so that it must intersect @xmath23 .
this gives a contradiction .
we conclude that @xmath90 has anticanonical degree @xmath10 , so that @xmath528 by theorem [ cmsbk ] .
now if @xmath529 is a line intersecting @xmath23 in at least two points , and @xmath530 is its transform , then @xmath531 .
we deduce that the secant variety @xmath532 of @xmath23 is not the whole @xmath22 , and this implies that @xmath23 is degenerate ( see for instance ( * ? ? ? * theorem 3.4.26 ) ) .
finally one can check that since @xmath3 is fano , the degree @xmath35 of @xmath23 is at most @xmath1 , completing the proof .
the proof of lemma [ contraction ] below depends on the following result , which is of independent interest .
[ scb ] let @xmath3 and @xmath112 be smooth connected projective varieties , and let @xmath533 be a @xmath41-bundle .
let @xmath4 be a locally unsplit dominating family of rational curves on @xmath3 with @xmath132=\dim(x)\ge 3 $ ] .
set @xmath534 .
then @xmath535 , and @xmath4 is the family of lines in the @xmath536 s .
let @xmath8 be a general point in @xmath3 , @xmath537 an irreducible component of @xmath7 , and @xmath538 the universal family .
let @xmath539 be the evaluation morphism , and @xmath540 its stein factorization .
then @xmath237 is a normal generalized cone ; in particular @xmath541 by ( * ? ? ?
* corollary ii.4.21 ) .
we claim that the composite map @xmath542 is finite ; in particular , it is dominant
. suppose otherwise .
then , by ( * ? ? ?
* corollary ii.4.21 ) , @xmath157 sends every curve in @xmath543 to a point .
thus @xmath132=2 $ ] , yielding a contradiction .
set @xmath544 , with natural morphisms @xmath545 , and @xmath546 .
@xmath547^{\nu}\ar[d]_{\tau } & x\ar[d]^{\pi}\\ t\ar[r]&y } \ ] ] notice that @xmath548 is a @xmath41-bundle .
let @xmath549 be the section of @xmath548 induced by @xmath550 ; we have @xmath551
. then @xmath552 , where @xmath553 is a rank 2 vector bundle on @xmath237 that fits into a short exact sequence @xmath554 with @xmath468 a line bundle on @xmath237 .
moreover , @xmath555 corresponds to the section @xmath556 , and @xmath557 identifies with the tautological line bundle , so that @xmath558 .
we prove that @xmath559 \le n$ ] .
suppose otherwise , namely that @xmath560>0 $ ] .
we have : @xmath561 and therefore @xmath562 let @xmath103\in v'_x$ ] , and let @xmath341 be an irreducible component of @xmath563 contained in @xmath556 . then , by the projection formula
: @xmath564 where @xmath503 is such that @xmath565 .
this implies that @xmath566 .
let now @xmath567\in v$ ] be a general point , and @xmath568 an irreducible component of @xmath569 not contained in @xmath556 . then
, as above , we must have @xmath570 .
on the other hand , @xmath571 since @xmath572 , and @xmath573 because @xmath541 .
this implies that @xmath574 yielding a contradiction . therefore
\le n$ ] .
let @xmath103\in v$ ] be a general point , and set @xmath575 .
then we must have : @xmath576 let @xmath577 be the irreducible component of @xmath578 passing through @xmath579 .
let @xmath580 be a general point , and set @xmath581 .
then @xmath582 since @xmath583 is dominant , and hence @xmath584=\dim ( v_y)_{y } + 2 \ge \dim y + 1\ ] ] by ( * ? ? ?
* theorems ii.1.7 and ii.2.16 ) .
thus @xmath585 , @xmath586 , and @xmath587=0.\ ] ] let @xmath588 be the normalization of @xmath589 , and let @xmath590 be the normalization of @xmath579 .
let @xmath183 be the transform of @xmath105 in @xmath588 . by the projection formula
, we have @xmath591 .
this implies that @xmath592 , and that @xmath183 is a fiber of @xmath593 .
in particular , @xmath594 is birational .
thus , @xmath577 is a locally unsplit family of rational curves with @xmath595=n=\dim(y)+1 $ ] .
therefore @xmath596 by theorem [ cmsbk ] , and @xmath597 is a section of @xmath157 . since @xmath560=0 $ ] , we must have @xmath598 , and hence @xmath599 . finally , @xmath600 , since @xmath601 .
this completes the proof of the lemma .
[ contraction ] let @xmath3 be a fano manifold of dimension @xmath16 and picard number @xmath24 , and suppose that @xmath3 has a locally unsplit dominating family of rational curves @xmath4 of anticanonical degree @xmath1 .
let @xmath145 be an irreducible component of @xmath513 for a general point @xmath6 .
then one of the following holds : 1 .
@xmath3 is the blow - up of @xmath22 along a linear subspace of codimension @xmath9 ; 2 .
@xmath602 ; 3 .
@xmath218 is an extremal ray of @xmath3 , whose associated contraction is divisorial and sends its exceptional divisor to a point ; the other extremal contraction of @xmath3 is either a ( singular ) conic bundle , or the blow - up of a smooth variety along a smooth subvariety of codimension @xmath9 . by remark [ locuns ] , @xmath603 $ ] .
this implies that @xmath95 is nef since @xmath604\geq 0 $ ] .
let @xmath114 be an extremal ray of @xmath3 such that @xmath147 , and let @xmath46 be the contraction of @xmath114 .
by lemma [ uno ] , @xmath112 is a smooth fano variety , and either @xmath148 is a blow - up with center @xmath146 smooth of codimension 2 , or @xmath148 is a conic bundle . set @xmath202 .
then @xmath203 is an extremal ray of @xmath3 by proposition [ divisorial ] .
suppose that we are in case @xmath169 of proposition [ divisorial ] .
we show that we are in case @xmath605 or @xmath606 .
the contraction of @xmath203 is @xmath607 , @xmath95 is a fiber , and hence @xmath604=0 $ ] . let @xmath163 be a general fiber of @xmath122 .
then the family of rational curves from @xmath4 contained in @xmath163 is locally unsplit with anticanonical degree @xmath608 , thus @xmath609 by theorem [ cmsbk ] .
let @xmath610 be a dense open subset such that @xmath611 is smooth . by tsen
s theorem , there exists a divisor @xmath612 on @xmath613 such that @xmath614 .
let @xmath365 be the closure of @xmath612 in @xmath3 .
then @xmath365 is @xmath122-ample since @xmath122 is elementary .
if @xmath615 , then @xmath616\cdot h^{n-1}=1 $ ] .
therefore , all fibers of @xmath122 are integral , and by ( * ? ? ?
* corollary 5.4 ) we have @xmath617 with @xmath123 a vector bundle of rank @xmath1 on @xmath41 . since @xmath3 is fano , it is not difficult to see that either @xmath618 , or @xmath3 is the blow - up of @xmath22 along a linear subspace of codimension @xmath9 . thus we get @xmath605 or @xmath606 .
case @xmath168 of proposition [ divisorial ] is @xmath619 .
we show that case @xmath193 of proposition [ divisorial ] does not occur .
suppose otherwise .
then the contraction @xmath122 of @xmath203 is small , it has a flip @xmath207 , @xmath208 is smooth , and there is a smooth @xmath41-bundle @xmath209 .
let @xmath103\in v$ ] be a general point .
then @xmath620 ( see ( * ? ? ?
* proposition ii.3.7 ) ) .
let @xmath621 be the irreducible component of @xmath622 passing through the transform @xmath183 of @xmath105 in @xmath208 .
we show that @xmath621 is a locally unsplit dominating family of rational curves on @xmath208 .
let @xmath623 be a general point .
if @xmath624 , then @xmath127\in s$ ] since @xmath514 $ ] and @xmath222 , and hence @xmath625 , yielding a contradiction
. therefore , @xmath626 , and hence @xmath627 is proper .
this proves that @xmath621 is a locally unsplit family of rational curves .
notice also that @xmath628=n$ ] . by lemma [ scb ] , @xmath629 , so that @xmath208 does not have small contractions , a contradiction .
this completes the proof of the lemma . finally , we prove the main result of this section .
[ prop_main ] let @xmath3 be a fano manifold of dimension @xmath16 , and suppose that @xmath3 has a locally unsplit dominating family @xmath4 of rational curves of anticanonical degree @xmath1 .
then @xmath52 .
if moreover @xmath33 , then @xmath3 is isomorphic to one of the varieties described in example [ ex ] , and @xmath127\equiv c_{{\widehat}{g}}+(d - a)f$ ] , where @xmath163 is a fiber of @xmath148 , and @xmath347 is the transform of a line in @xmath630 ( notations as in example [ ex ] ) .
_ let @xmath145 be an irreducible component of @xmath513 for a general point @xmath6 .
then @xmath603 $ ] by remark [ locuns ] , and by lemma [ uno ] we have @xmath52 .
notice that @xmath95 is nef , because @xmath604\geq 0 $ ] .
we assume that @xmath33 . by theorem [ main2 ] ,
@xmath3 is as described in example [ ex2 ] .
we use the notations of example [ ex2 ] and theorem [ main2 ] ; in particular we refer the reader to table [ table ] for the intersection table of @xmath3 .
we have to show that @xmath631 , that @xmath381 , and that @xmath127\equiv c_g+a{\widehat}{f}\equiv c_{{\widehat}{g}}+(d - a)f$ ] ( see ) .
[ secondpart ] notice that @xmath632=0 $ ] .
indeed if for instance @xmath633 , as @xmath384 , then there exists @xmath179 such that @xmath127=\lambda [ c_g]$ ] . on the other hand
@xmath632\geq 0 $ ] while @xmath634 , and we conclude that @xmath632=0 $ ] .
set @xmath635 .
since @xmath636,[{\widehat}{f}],[c_g]$ ] are a basis of @xmath98 , and we can write : @xmath637\equiv\alpha f+\beta{\widehat}{f}+\gamma c_g,\ ] ] with @xmath638 . intersecting with @xmath205 yields @xmath639 , and intersecting with @xmath640 yields @xmath641 : @xmath642 } [ v]\equiv(n - m\gamma ) f+a\gamma\delta{\widehat}{f}+\gamma c_g.\ ] ] notice that @xmath643 , because @xmath636 $ ] belongs to the extremal ray @xmath114 , and @xmath127\not\in r$ ]
moreover @xmath644=\gamma\delta(d - a)$ ] , so we obtain @xmath645 . finally , we also have @xmath646=n - m\gamma\geq 0 $ ] . [ base ] by we have @xmath647=-k_z\cdot \gamma c_z=\gamma m\leq n=\dim z+1.\ ] ] let @xmath103\in v$ ] be a general point , and let @xmath648 be the irreducible component of @xmath649 passing through @xmath650 .
let @xmath580 be a general point , and set @xmath651 .
then @xmath652 , and hence @xmath653 = \dim ( v_z)_{z}+2\ge \dim z + 1\ ] ] by ( * ? ? ?
* theorems ii.1.7 and ii.2.16 ) . thus using ( [ degree ] )
we get @xmath654 , @xmath655 , @xmath656 , and @xmath657 = \dim z + 1.\ ] ] therefore becomes : @xmath658\equiv \frac{n}{m}\left(a\delta{\widehat}{f}+c_g\right).\ ] ] we show that @xmath648 is a locally unsplit family of rational curves on @xmath109 . as @xmath656 , and @xmath7 is proper , the curves in @xmath659 and their possible degenerations are images of curves in @xmath7 , which are integral . therefore , in order to show that @xmath648 is locally unsplit , we only have to exclude the possibility that @xmath122 has degree @xmath660 on some curve of @xmath7 .
we proceed by contradiction , and assume that for some @xmath661 \in v_x$ ] , the morphism @xmath662 has degree @xmath663 .
let @xmath664 be the normalization , set @xmath665 , and denote by @xmath666 ( respectively , @xmath667 ) the natural morphisms .
@xmath668^{\nu}\ar[d]_{\pi_{|\mathbb{f } } } & { y}\ar[d]_{\pi } & x\ar[l]_{\sigma}\ar[dl]^{{\varphi}}\\ { \ell}\ar[r]&{z } & } \ ] ] let @xmath669 be the transform of @xmath670 .
let moreover @xmath671 be a minimal section of @xmath672 , and @xmath673 a general fiber .
we have @xmath674=0 $ ] by [ secondpart ] , and @xmath675 because @xmath676 is general , hence @xmath677 .
the pull - back of @xmath309 to @xmath588 is precisely @xmath678 , so that @xmath679
. moreover @xmath672 has degree @xmath680 on @xmath681 .
set @xmath682 .
then @xmath683 , and we obtain : @xmath684 consider now @xmath685 .
this is a section of @xmath686 disjoint from @xmath678 , so that @xmath687 .
we also have @xmath688 , and @xmath646=0 $ ] by , so using the projection formula : @xmath689 since @xmath690 , we have @xmath691 . moreover @xmath692 because @xmath676 , hence @xmath693 , and @xmath694 is a zero - dimensional scheme .
we see @xmath694 as a zero - dimensional subscheme of @xmath695 ; in particular , it is determined by its support and by its multiplicity at each point .
we write @xmath696 .
let @xmath697 be the blow up of @xmath588 along @xmath694 ; note that there is a natural morphism @xmath698 : @xmath699^{{\widetilde}{\nu}}\ar[d]_{\varpi } & x\ar[d]_{\sigma}\ar@/^1pc/[dd]^{{\varphi } } \\ { \mathbb{f}}\ar[r]^{\nu}\ar[d]_{\pi_{|\mathbb{f } } } & { y}\ar[d]_{\pi}\\ { \ell}\ar[r]&{z } } \ ] ] observe that @xmath700 is a normal surface , with l.c.i singularities .
indeed , locally over @xmath701 , @xmath700 can be described as the blow - up of @xmath702 at the ideal @xmath703 . in particular
, @xmath700 is smooth at @xmath704 if @xmath705 , otherwise it has a du val singularity of type @xmath706 .
set @xmath707 ( with reduced scheme structure ) ; then @xmath708 let @xmath709 be the transform of @xmath710 .
then @xmath711 , and using , , and , we get : @xmath712 the curve @xmath713 is an integral rational curve in @xmath714 , of anticanonical degree @xmath715 by .
we fix a point @xmath716 such that @xmath717 . by (
* theorems ii.1.7 and ii.2.16 ) , every irreducible component of @xmath718 through @xmath719 $ ] has dimension @xmath720 , because @xmath663 by assumption .
let @xmath721 be an irreducible curve through @xmath722 $ ] .
if @xmath399 is proper , we get @xmath723 and hence @xmath724 by ( * ? ? ?
* corollary iv.4.21 ) , a contradiction .
if instead @xmath399 is not proper , the closure of its image in @xmath725 contains points corresponding to non - integral curves . but this gives again a contradiction , because @xmath7 is proper .
we conclude that @xmath648 is a locally unsplit family of rational curves on @xmath109 , with @xmath726=\dim z+1 $ ] by .
hence @xmath631 by theorem [ cmsbk ] , @xmath337 is a line , @xmath727 , @xmath728 , and @xmath729 .
finally @xmath127\equiv a{\widehat}{f}+c_g$ ] by .
let @xmath3 be one of the varieties introduced in example [ ex ] ; we use the same notations as in examples [ ex ] and [ ex2 ] , with @xmath383 .
we construct a dominating family of rational curves on @xmath3 , and then show that it is locally unsplit .
notice that the condition @xmath381 is necessary to ensure the existence of such a family , see [ secondpart ] .
we first consider the case @xmath58 , so that @xmath730 , and @xmath3 is the blow - up of @xmath731 along @xmath732 , where @xmath733 is a fixed point .
the general curve of the family @xmath4 is the transform in @xmath3 of @xmath734 , where @xmath735 and @xmath736 is a line .
therefore @xmath4 is a locally unsplit dominating family of rational curves , and for a general point @xmath6 , @xmath7 is isomorphic to the variety of lines through a fixed point in @xmath536 ; hence @xmath59 .
suppose from now on that @xmath60 , and set @xmath737 let @xmath738 be a line not contained in @xmath23 , and let @xmath739 be such that @xmath740 .
set @xmath741 , and denote by @xmath686 the restriction of @xmath157 to @xmath588 .
we have @xmath742 the minimal section of @xmath686 , and @xmath743 is another section with @xmath744 , where @xmath745 is a fiber of @xmath686 ( see figure [ fig2 ] ) . since @xmath746 is a hypersurface of degree @xmath35 , and @xmath747 , @xmath748 is a zero - dimensional scheme of length @xmath35 .
moreover @xmath749 is isomorphic to @xmath748 , because @xmath319 is a section of @xmath157 . in particular @xmath749
can be seen as a closed subscheme of @xmath750 , hence it is determined by its support and by its multiplicity at each point .
recall that @xmath751 by assumption .
let us consider a closed subscheme @xmath90 of @xmath749 , of length @xmath62 .
again , @xmath90 is determined by its support and by its multiplicity at each point , so that without loss of generality we can write @xmath752 , where @xmath701 are possibly equal points in @xmath753 ( and each @xmath701 appears at most @xmath754 times , if @xmath754 is the multiplicity of @xmath749 at @xmath701 ) .
[ construction ] we associate to @xmath755 a smooth rational curve @xmath128 , of anticanonical degree @xmath1 , and containing @xmath8 . set @xmath756 .
we write @xmath757 ( respectively , @xmath758 ) for the ideal sheaf of @xmath90 ( respectively , @xmath759 ) in @xmath588 . we claim that @xmath760 and that the corresponding curve on @xmath588 is smooth and irreducible . since @xmath761 , we must have @xmath762 .
let @xmath763 be a curve containing @xmath90 and @xmath759 .
observe first that @xmath764 is irreducible .
otherwise , since @xmath765 , there is a unique irreducible component @xmath766 of @xmath764 such that @xmath767 and @xmath768 for some @xmath769 ( @xmath766 is a section of @xmath686 ) .
in particular , @xmath770 with @xmath771 . by ( * ? ? ? * corollary v.2.18 ) , we have @xmath772 and @xmath773 .
thus @xmath774 where the @xmath775 s are possibly equal fibers of @xmath686 .
notice that @xmath90 is a subscheme of both @xmath774 and @xmath776 . since @xmath776 is disjoint from @xmath678
, we must have @xmath777 on the other hand @xmath319 intersects transversally any fiber of @xmath686 , thus we get @xmath778 , and up to renumbering we can assume that @xmath775 is the fiber of @xmath686 containing @xmath701 .
this implies that @xmath779 , which contradicts our choice of @xmath759 . thus @xmath764 is irreducible ,
hence it is a section of @xmath686 , and @xmath780 . to show that @xmath764 is unique ,
let @xmath781 be another curve containing @xmath90 and @xmath759 .
then @xmath782 is irreducible , @xmath783 , and @xmath784 , which implies that @xmath785 .
this show our claim .
we remark that : @xmath786 indeed @xmath764 is not contained in @xmath319 , because @xmath787 . moreover @xmath788 and @xmath789 , so that @xmath790 .
we define @xmath128 to be the transform of @xmath791 , so that @xmath105 is a smooth rational curve through @xmath8 .
it is not difficult to see that @xmath792 , and hence @xmath793 ( see table [ table ] ) .
[ free ] if @xmath91 is either reduced , or has a unique non reduced point of multiplicity @xmath9 , then : @xmath794 hence @xmath105 is a standard , free , smooth rational curve .
suppose first that @xmath795 .
this implies that @xmath122 is smooth in a neighbourhood of @xmath105 , thus the map @xmath796 is onto .
its kernel is a torsion free sheaf of rank 1 on @xmath105 , hence a locally free sheaf , of degree @xmath797 .
since @xmath798 , this implies the statement .
suppose now that @xmath799 , and let @xmath800 be a point in this intersection .
set @xmath801 , and let @xmath802 ( respectively , @xmath803 ) be the multiplicity of @xmath804 in @xmath91 ( respectively , in @xmath805 ) .
notice that @xmath348 and @xmath349 are cartier divisors on @xmath3 , and they do not contain @xmath105 in their support . as @xmath148 is an isomorphism on @xmath105 , we have @xmath806 . on the other hand : @xmath807
therefore we have the following equalities of divisors on @xmath105 : @xmath808 in particular @xmath804 has multiplicity @xmath809 in @xmath810 , and since @xmath811 , we have @xmath812 .
we deduce that @xmath813 , @xmath814 , and @xmath804 is the unique non - reduced point in @xmath91 .
as @xmath815 , we also deduce that @xmath105 intersects @xmath816 only in @xmath800 . moreover , since @xmath105 is transverse to @xmath348 ( and to @xmath349 ) in @xmath800 , we have : @xmath817 now let us consider the tangent morphism @xmath818 , which is surjective outside @xmath819 , and has rank @xmath20 in every point @xmath820 , with image @xmath821 .
let us also consider the natural surjective morphism : @xmath822 since @xmath348 and @xmath349 intersect transversally , a local computation shows that the image of @xmath823 is the subsheaf of @xmath824 given by the kernel of @xmath825 . in other words , we have an exact sequence : @xmath826 let us now restrict to the curve @xmath105 .
then @xmath827 is a skyscraper sheaf , and we have an exact sequence : @xmath828 where @xmath829 is just the evaluation map .
we have @xmath830 , and the factor @xmath831 is contained in @xmath832 , because @xmath833 .
therefore we get an induced surjective morphism @xmath834 , whose kernel is @xmath835 .
this gives an exact sequence @xmath836 which yields @xmath837 .
thus @xmath838 yields a surjective map @xmath839 .
the kernel of this morphism is a torsion free sheaf of rank 1 on @xmath105 , hence a locally free sheaf , of degree @xmath840 .
finally , the exact sequence @xmath841 gives the statement .
[ lu ] let @xmath183 be an effective one - cycle in @xmath3 such that @xmath842 and @xmath843 .
then @xmath183 is integral and is obtained as in [ construction ] , for some choices of @xmath844 , @xmath845 , and @xmath846 . in particular
, @xmath183 is again a smooth , connected rational curve . since @xmath847
, it follows that @xmath848 is a line @xmath849 in @xmath536 , and @xmath850 because @xmath851 .
moreover , there is a unique irreducible component @xmath852 of @xmath183 that maps onto @xmath849 , and @xmath853 is a birational morphism .
therefore we have @xmath854 where the @xmath855 s are ( possibly equal ) fibers of @xmath856 , the @xmath857 s are fibers of @xmath858 , and the @xmath859 s are fibers of @xmath122 over @xmath860 .
moreover @xmath861 , and using table [ table ] we get : @xmath862 since @xmath852 is irreducible and @xmath863 , the intersections @xmath864 and @xmath865 can not be both negative ; this yields @xmath866 .
then @xmath852 can not be contained in @xmath867 , because @xmath183 is not contained in @xmath868 .
we conclude that @xmath869 , and @xmath870 is integral .
set @xmath871 , and denote by @xmath872 the restriction of @xmath157 to @xmath873 .
we consider the curve @xmath874 . since @xmath875 has degree one , we have @xmath876 , where @xmath877 is the minimal section of @xmath878 ,
@xmath879 is one of its fibers , and @xmath429 .
on the other hand we have @xmath880 where @xmath881 .
therefore @xmath431 , and @xmath882 .
finally , as @xmath883 and @xmath766 is smooth , @xmath766 intersects @xmath47 in a zero - dimensional subscheme @xmath884 of length @xmath62 .
this shows that @xmath183 is a curve obtained via the construction described in [ construction ] . if @xmath58 , the statement is clear .
suppose that @xmath60 . by theorem [ grass ] ,
pairs @xmath885 where @xmath736 is a line not contained in @xmath23 , and @xmath886 is a subscheme of length @xmath62 , vary in an irreducible family of dimension @xmath887 . using this , it is not difficult to show that varying @xmath888 , construction [ construction ] yields an irreducible algebraic family of smooth , connected rational curves in @xmath3 , of anticanonical degree @xmath1 .
more precisely , we get a locally closed irreducible subvariety @xmath889 of @xmath5 of dimension @xmath890 , whose points correspond to curves @xmath105 obtained as in [ construction ] . by lemma
[ free ] , a general curve @xmath105 from @xmath889 is free and hence yields a smooth point in @xmath5 .
moreover , @xmath891 } { \operatorname{ratcurves}^n}(x)=-k_x\cdot [ v]+n-3=2n-3 $ ] .
this implies that the closure @xmath4 of @xmath889 in @xmath5 is an irreducible component , and that @xmath4 gives a dominating family of curves of anticanonical degree @xmath1 . on the other hand , lemma [ lu ]
shows that @xmath7 is proper for every @xmath739 , so that @xmath4 is a locally unsplit family .
[ minimaldegree ] let @xmath502 be an ample line bundle on @xmath3 .
if @xmath892 , then @xmath4 is not a dominating family of rational curves of minimal degree with respect to @xmath502 .
indeed , the family @xmath90 of curves on @xmath3 whose points correspond to smooth fibers of @xmath122 is a locally unsplit dominating family of rational curves , and @xmath87\equiv f+{\widehat}{f}$ ] .
thus @xmath893=\frac{1}{2}\left(\mathscr{h}\cdot c_g+\mathscr{h}\cdot c_{{\widehat}{g}}+ a \mathscr{h}\cdot { \widehat}{f}+(d - a ) \mathscr{h}\cdot f\right ) > \mathscr{h}\cdot [ w].\ ] ] the first part of the statement follows from proposition [ prop_main ] .
assume that @xmath33 .
then , again by proposition [ prop_main ] , @xmath3 is isomorphic to one of the varieties described in example [ ex ] , and @xmath127\equiv c_{{\widehat}{g}}+(d - a)f$ ] ( notation as in example [ ex ] and proposition [ prop_main ] ) . by , this is the same as @xmath127\equiv c_g+a{\widehat}{f}$ ]
this means that the curves of the family @xmath4 are numerically equivalent to the curves obtained in construction [ construction ] .
thus lemma [ lu ] implies that the family @xmath4 coincides with the family of curves constructed in the proof of proposition [ examples ] . as before
, we can assume that @xmath60 .
let @xmath739 be a general point , and set @xmath894 .
let @xmath55 be the morphism of degree @xmath35 induced by the linear projection @xmath56 from @xmath57 , where we see @xmath26 as the variety of lines @xmath579 through @xmath57 in @xmath536 .
then the pairs @xmath885 such that @xmath736 is a line through @xmath8 , and @xmath895 is a zero - dimensional subscheme of length @xmath62 , are parametrized by the relative hilbert scheme @xmath896}(a/\mathbb{p}^{n-2})$ ] . since @xmath8 is general , @xmath897 is integral by theorem [ genus_hilbert ] .
therefore , using ( * ? ? ?
* corollary iii.12.9 ) , construction [ construction ] can be made in family over @xmath898 .
one first constructs a subscheme @xmath899 , such that the fiber of @xmath900 over @xmath885 is the curve @xmath764 . as @xmath901 intersects @xmath902 along the cartier divisor @xmath903 ( see ) , the transform @xmath904 of @xmath899 is isomorphic to @xmath901 . in particular
, the fiber of @xmath905 over @xmath885 is the curve @xmath105 .
as @xmath105 is a smooth rational curve through @xmath8 , @xmath905 is a @xmath41-bundle with a section @xmath906 , given by @xmath907 . in particular
, @xmath908 is the projectivization of a rank @xmath9 vector bundle over @xmath898 , and it is locally trivial in the zariski topology .
fix a point @xmath131 .
let @xmath909 be an open subset such that @xmath910 .
we can assume that the section @xmath911 , under this isomorphism , is identified with the constant section @xmath912 .
therefore we get a morphism over @xmath612 : @xmath913 which is an embedding on @xmath914 , and sends @xmath915 to @xmath916 , for every @xmath917 .
this yields a morphism @xmath918 .
since @xmath8 is general , every curve in @xmath7 is free , and @xmath130 contains a union of smooth irreducible components @xmath919 , whose image in @xmath920 is @xmath7 . by construction ,
the morphism @xmath918 takes values in @xmath919 . by (
* theorem ii.2.16 ) , these morphisms glue together and yield a morphism @xmath921 by lemma [ lu ] the morphism @xmath922 is surjective .
moreover , the pair @xmath885 determines uniquely the curve @xmath105 in construction [ construction ] . as
@xmath105 is smooth , it corresponds to a unique point in @xmath7 ; therefore @xmath922 is injective .
finally , @xmath7 being smooth , we conclude that @xmath922 is an isomorphism , @xmath923 is smooth , and @xmath7 is irreducible .
notice that @xmath62 and @xmath382 yield not only the same variety @xmath3 ( see remark [ symm ] ) , but also the same family @xmath4 of rational curves on @xmath3 , and @xmath924}(a/\mathbb{p}^{n-2})\cong{\operatorname{hilb}}^{[d - a]}(a/\mathbb{p}^{n-2})$ ] for general @xmath6 .
the cases @xmath925 and @xmath926 are the simplest ones . for the reader
s convenience , we describe explicitly the latter .
[ d = a+1 ] if @xmath927 , then @xmath928 is the blow - up of @xmath22 at a point .
thus @xmath3 is the blow - up of @xmath22 along @xmath929 , where @xmath23 is smooth , of dimension @xmath20 , degree @xmath35 , contained in a hyperplane @xmath365 , and @xmath930 .
this is one of the few examples of fano manifolds obtained by blowing - up a point in another manifold , see @xcite .
the general curve of the family @xmath4 is the transform in @xmath3 of a line in @xmath22 intersecting @xmath23 in one point .
therefore for a general point @xmath6 we have @xmath931}(a/\mathbb{p}^{n-2})$ ] . from the point of view of the family of curves
, this is essentially the same example as example [ examplem ] ; see also ( * ? ? ?
* example 1.7 ) .
let us consider now the morphism @xmath932 where @xmath8 is a general point ( notation as in the introduction ) .
we recall the following useful observation .
[ standard ] let @xmath103\in v_x$ ] .
then @xmath66 is an immersion at @xmath103 $ ] if and only if @xmath105 is standard , see ( * ? ? ?
* proposition 1.4 ) and ( * ? ? ?
* proposition 2.7 ) . as @xmath8 is general , @xmath933 is smooth at @xmath8 , and the differential of @xmath122 at @xmath8 induces the linear projection @xmath934 from the point @xmath935\in \mathbb{p}(t_{x , x}^*)$ ] .
we have @xmath935\not\in\mathcal{c}_x$ ] , because @xmath936 is an isomorphism for every @xmath105 in @xmath7 , and the projection restricts to a morphism @xmath937 .
observe that there is a commutative diagram : @xmath938^{\phi}\ar[r]_{\psi}\ar[dr]_{\pi } & { v_x } \ar[r]_{\tau_x } & { \mathcal{c}_x}\ar[dl]^{\pi'}\\ & { { \mathbb{p}}^{n-2 } } & } \ ] ] where we have set @xmath939 .
since @xmath89 has degree @xmath70 , @xmath922 is an isomorphism , and @xmath66 is birational , we conclude that @xmath940 has degree @xmath70 , therefore @xmath941 is a hypersurface of degree @xmath70 .
moreover @xmath65 is irreducible , because @xmath7 is .
we have seen in the above example that @xmath66 is an isomorphism for @xmath927 and @xmath942 , and the statement is clear if @xmath58 or @xmath378 .
we suppose from now on that @xmath73 ; notice that in particular @xmath943 . since @xmath922 is an isomorphism , the statement follows if we show that the closed subset where @xmath944 is not an isomorphism ( respectively , an immersion ) has codimension @xmath74 ( respectively , @xmath9 ) .
let @xmath945 be a general line .
then @xmath946 is a smooth plane curve of degree @xmath35 , and @xmath947}(a_l / l)$ ] is a smooth curve of genus @xmath948 by theorem [ genus_hilbert ] ( 4 ) . on the other hand
, @xmath949 is a plane curve of degree @xmath70 , hence it has arithmetic genus @xmath950 .
we claim that @xmath951 can not be an isomorphism onto its image . by contradiction , if it were , the two curves should have the same genus , so we get : @xmath952 this yields easily @xmath953 , or equivalently @xmath954 .
since @xmath955 , we must have @xmath956 and hence @xmath957 equivalently , we get @xmath958 .
it is easy to see that this contradicts the assumption @xmath73 .
we conclude that the closed subset where @xmath944 is not an isomorphism has codimension @xmath74 .
let us consider again the plane curve @xmath959 . since @xmath960 is a general projection , every non - reduced fiber contains just one double point .
hence , for every @xmath88\in l$ ] , the intersection @xmath91 is either reduced , or has a unique non - reduced point , of multiplicity @xmath9 . by lemma
[ free ] , for every @xmath886 , the corresponding curve @xmath128 is standard , therefore @xmath66 is an immersion at @xmath961)$ ] by remark [ standard ] .
we have shown that @xmath944 is an immersion in every point of @xmath962 , hence the closed subset where @xmath944 is not an immersion has codimension at least @xmath9 .
suppose now that @xmath963 , and let @xmath964 be a general plane .
then @xmath965 is a smooth surface of degree @xmath35 in @xmath966 , and @xmath967 is the projection from a general point . let @xmath968 be the branch curve of this projection .
it is classically known that @xmath399 has only nodes and cusps , see @xcite . moreover ,
the fiber of @xmath967 over a smooth point @xmath969 contains just one double point , the fiber over a node contains two double points , and the fiber over a cusp contains one triple point ( see ( * ? ? ?
* proposition 3.7 ) ) .
the number of nodes in @xmath399 depends only on the degree @xmath35 of @xmath970 , and more precisely it is @xmath971 , see ( * ? ? ?
* lemma 3.2(a ) ) . since @xmath943
, @xmath399 contains nodes , and we conclude that there is at least one @xmath88\in p$ ] such that @xmath972 . as @xmath973
, we can consider the subscheme @xmath974 and the point @xmath87\in \mathcal{h}_z$ ] .
notice that the points @xmath975 and @xmath976 do appear in @xmath90 , because @xmath977 .
then by theorem [ genus_hilbert ] ( 2 ) , the tangent space of the fiber @xmath978)$ ] at @xmath87 $ ] has dimension @xmath9 . on the other hand , as @xmath940 is a linear projection from a point , the tangent space of the fiber @xmath979)$ ] at @xmath980)$ ] has dimension at most @xmath74 .
this shows that @xmath944 is not an immersion at @xmath87 $ ] , and hence that the closed subset where @xmath944 is not an immersion has codimension @xmath9 .
let @xmath3 and @xmath4 be as in proposition [ examples ] .
assume that @xmath73 , and that @xmath963 .
then non - standard curves of the family @xmath4 cover @xmath3 .
this follows from remark [ standard ] and theorem [ tau_x ] .
let @xmath3 and @xmath4 be as in proposition [ examples ] , and consider the case @xmath981 .
then @xmath3 is fano if and only if @xmath982 and @xmath983 , in particular @xmath984 .
thus the only case where @xmath3 is fano and @xmath66 is not an isomorphism is for @xmath985 and @xmath986 .
this fano @xmath2-fold @xmath3 is n. 9 in @xcite . in this case
@xmath987 is a smooth quartic , @xmath988}(a/{\mathbb{p}}^1)$ ] is a smooth connected curve of genus @xmath989 , and @xmath990 has degree @xmath991 . using theorem [ genus_hilbert ] ( 2 )
, one can describe precisely the ramification of @xmath89 .
on the other hand , @xmath992 is an irreducible curve of degree @xmath991 and arithmetic genus @xmath993 .
the normalization @xmath994 is an immersion , but it is not injective .
the proof of theorem [ v_x ] relies on the following results of independent interest .
[ genus_hilbert ] fix integers @xmath77 , @xmath62 , and @xmath35 , such that @xmath78 and @xmath79 .
let @xmath80 be a smooth hypersurface of degree @xmath35 , @xmath995 a general point , and @xmath996 the linear projection from @xmath57 ( where we see @xmath83 as the variety of lines through @xmath57 in @xmath84 ) . 1 .
the relative hilbert scheme @xmath61}(a/\mathbb{p}^{m})$ ] is an integral local complete intersection scheme of dimension @xmath77 , and the natural morphism @xmath997}(a/\mathbb{p}^{m})\to \mathbb{p}^{m}\ ] ] is flat and finite of degree @xmath70 .
2 . let @xmath88\in{\mathbb{p}}^m$ ] and @xmath87\in\pi^{-1}([\ell])$ ] .
write @xmath998 with @xmath999 and @xmath1000 for @xmath1001 , and @xmath1002 with @xmath1003 .
the tangent space of the fiber @xmath978)$ ] at @xmath87 $ ] has dimension @xmath1004 .
@xmath89 is smooth at @xmath87 $ ] if and only if @xmath90 is a union of irreducible components of @xmath91 , equivalently : @xmath1005 .
4 . suppose that @xmath1006 .
then the curve @xmath61}(a/\mathbb{p}^1)$ ] is smooth of genus @xmath1007 .
[ local ] let @xmath1008 be a positive integer , and set : @xmath1009}{(t^{h})}\quad\text { and } \quad f:={\operatorname{spec}}\lambda.\ ] ] let @xmath90 be a non - empty closed subscheme of @xmath163 , with ideal @xmath1010 , where @xmath1011 is an integer .
then : 1 . @xmath1012 ; 2 .
@xmath1013 has dimension zero , @xmath1014 , and the tangent space of @xmath1013 at @xmath87 $ ] has dimension @xmath1015 ; 3 .
@xmath1013 is smooth at @xmath87 $ ] if and only if @xmath1016 .
we have a short sequence of @xmath1017-modules : @xmath1018 where the second morphism is given by @xmath1019 , so that @xmath1020 as @xmath1017-modules . using this isomorphism
, a direct computation shows that @xmath1021 . applying the functor @xmath1022 to the above sequence , and using the vanishing of @xmath1023
, we get the exact sequence of @xmath1017-modules : @xmath1024 similarly as before , observe that @xmath1025 as @xmath1017-modules . moreover , if @xmath1026 is the quotient map , it is easy to see that @xmath1027 is an isomorphism of @xmath1017-modules
. therefore @xmath1028 and we obtain ( 1 ) : @xmath1029 the hilbert scheme of @xmath163 is supported on finitely many points , thus it has dimension zero .
recall that by ( * ? ? ?
* definition i.2.6 ) , the obstruction space @xmath1030 is a subspace of @xmath1031 . then ( 2 ) follows from ( 1 ) and
* theorems i.2.8.1 and i.2.8.4 ) .
[ local2 ] let @xmath1032 be a finite morphism between smooth quasi - projective varieties .
let @xmath163 be a fiber of @xmath31 , and @xmath1033 a reduced subscheme of length @xmath62 .
let @xmath85}(a / t)_{red}\to t$ ] be the natural morphism .
suppose that @xmath163 has a unique non reduced point at @xmath804 .
then the hilbert scheme @xmath1034}(a / t)_{red}$ ] is smooth at @xmath87 $ ] , and there exists a neighbourhood for the euclidean topology @xmath1035}(a / t)_{red}$ ] ( respectively , @xmath1036 ) of @xmath87 $ ] ( respectively , @xmath804 ) and an isomorphism @xmath1037 such that the diagram : @xmath1038_{p_{|u_1}}\ar[rr]^{\iota } & & u\ar[dl]^{\pi_{|u}}\\ & t & } \ ] ] commutes .
recall that the hilbert - chow morphism @xmath61}(a ) \to a^{(a)}:=a^a/\mathbb{s}_a$ ] maps a zero - dimensional subscheme of length @xmath62 of @xmath23 to the associated effective @xmath1039-cycle of degree @xmath62 .
notice that @xmath87 $ ] is contained in the open subset of @xmath1034}(a)$ ] where the hilbert - chow morphism is an isomorphism , and that @xmath1034}(a)$ ] is smooth at @xmath87 $ ] since @xmath87 $ ] is reduced .
let @xmath1040 be an open neighbourhood of @xmath1041 for the euclidean topology such that @xmath1042 , @xmath1043 if @xmath1001 , @xmath1036 is an open neighbourhood of @xmath804 , and @xmath1044 is an isomorphism for @xmath1045 .
then , up to shrinking @xmath4 if necessary , the morphism @xmath1046\in a^{(a)}\ ] ] yields a holomorphic map @xmath1047}(a / t)_{red}\subset \textup{hilb}^{[a]}(a).\ ] ] notice that @xmath1048}(a / t)_{red}$ ] is an immersion , and that it induces a bijective morphism @xmath1049 such that @xmath1050 .
our claim follows easily .
let @xmath1051 be a line passing through @xmath57 , and set @xmath1052 , so that @xmath163 is a zero - dimensional subscheme of @xmath1053 .
we have : @xmath1054)={\operatorname{hilb}}^{[a]}(f);\ ] ] in particular @xmath89 is a finite morphism , and @xmath1055}(a/\mathbb{p}^{m } ) \le m$ ] .
let @xmath87\in \textup{hilb}^{[a]}(a/\mathbb{p}^{m})$ ] be a point over @xmath88 \in \mathbb{p}^{m}$ ] .
applying lemma [ local ] to every connected component of @xmath978)$ ] , we get ( 2 ) and ( 3 ) , and also that @xmath1056 .
thus , by ( * ? ? ?
* theorems i.2.10.3 and i.2.10.4 ) , any irreducible component of @xmath61}(a/\mathbb{p}^{m})$ ] through @xmath87 $ ] has dimension @xmath77 , and @xmath89 is a local complete intersection morphism .
in particular , @xmath61}(a/\mathbb{p}^{m})$ ] is a local complete intersection scheme , and @xmath89 is a flat finite morphism . by ( 3 ) , @xmath89 is tale over @xmath88 $ ] if and only if @xmath91 is reduced ,
_ i.e. _ @xmath63 is tale over @xmath88 $ ] .
therefore @xmath61}(a/\mathbb{p}^{m})$ ] is generically smooth over @xmath1057 .
in particular , @xmath61}(a/\mathbb{p}^{m})$ ] is generically reduced and hence reduced as it is a cohen - macaulay scheme .
we proceed to show that the scheme @xmath61}(a/\mathbb{p}^{m})$ ] is irreducible .
this follows from the fact that since @xmath57 is general , the monodromy group of the projection @xmath1058 is the whole symmetric group @xmath1059 , see ( * ? ? ?
* proposition 2.3 ) .
let @xmath1060 be a dense open subset such that @xmath63 and @xmath89 are tale over @xmath1061 .
set @xmath1062 and @xmath1063}(a_0/u)$ ] .
since @xmath612 is dense in @xmath61}(a/\mathbb{p}^{m})$ ] , and @xmath612 is smooth , we are reduced to show that @xmath612 is connected .
notice that @xmath612 is a closed subscheme of @xmath1034}(a_0)$ ] , and that every @xmath87 $ ] in @xmath612 is a reduced subscheme of @xmath1064 , so that @xmath612 is contained in the open subset of @xmath1034}(a_0)$ ] where the hilbert - chow morphism @xmath61}(a_0 ) \to ( a_0)^{(a)}$ ] is an isomorphism .
let @xmath1065 $ ] and @xmath1066 $ ] be two points in @xmath612 that map to a given point in @xmath1061 .
since any irreducible component of @xmath612 maps onto @xmath1061 , it is enough to prove that there is a path in @xmath612 joining @xmath1065 $ ] to @xmath1066 $ ] .
let @xmath1067\in ( a_0)^a$ ] mapping to @xmath1068 $ ] .
since the monodromy group of @xmath63 is the symmetric group @xmath1059 , there is a path @xmath1069\to ( a_0)^a$ ] joining @xmath1070 $ ] to @xmath1071 $ ] , such that the points in @xmath1072 are distinct and contained in a fiber of @xmath63 , for every @xmath1073 $ ] .
this yields a path joining @xmath1065 $ ] to @xmath1066 $ ] in @xmath612 , and proves that @xmath61}(a/\mathbb{p}^{m})$ ] is integral .
let @xmath87\in \textup{hilb}^{[a]}(a/\mathbb{p}^1)$ ] be a point over @xmath88\in{\mathbb{p}}^1 $ ] . by ( 3 ) ,
either @xmath89 is tale at @xmath87 $ ] , or @xmath91 has a double point , @xmath90 is reduced , and contains this point in its support .
note that there are exactly @xmath1077 such @xmath87 $ ] s , and that @xmath89 has ramification index 2 at any of these points by lemma [ local2 ] .
if @xmath89 is tale at @xmath87 $ ] , then @xmath61}(a/\mathbb{p}^1)$ ] is obviously smooth at @xmath87 $ ] .
otherwise , @xmath61}(a/\mathbb{p}^1)$ ] is smooth at @xmath87 $ ] by lemma [ local2 ] .
this shows that @xmath61}(a/\mathbb{p}^1)$ ] is a smooth curve .
the same proof shows that for _ every _
@xmath1079 , the relative hilbert scheme @xmath61}(a/\mathbb{p}^{m})$ ] is a reduced local complete intersection scheme of dimension @xmath77 , and that the natural morphism @xmath85}(a/\mathbb{p}^{m})\to \mathbb{p}^{m}$ ] is flat and finite of degree @xmath70 .
moreover , ( 2 ) and ( 3 ) hold true .
we need also a slightly more general version of the previous construction , as follows .
let @xmath100 be a smooth hypersurface of degree @xmath36 , and fix @xmath1080 .
set @xmath1081\in g(1,m+1)\,|\,\ell\text { is not contained in $ a$}\right\},\ ] ] with its universal family @xmath1082,z)\in\mathcal{g}\times{\mathbb{p}}^{m+1}\,|\,z\in\ell\}$ ] .
let us consider the intersection : @xmath1083 the induced morphism @xmath1084 is finite and flat , of degree @xmath35 ; the fiber over a line @xmath88 $ ] is @xmath91 .
the relative hilbert scheme @xmath61}(\mathcal{i}/\mathcal{g})$ ] parametrizes pairs @xmath885 where @xmath1051 is a line not contained in @xmath23 , and @xmath90 is a subscheme of length @xmath62 of @xmath91 . with the same proof of theorem [ genus_hilbert ] , one shows : set @xmath1087 $ ] .
if @xmath58 , there is nothing to prove . if @xmath60 , consider the relative hilbert scheme @xmath61}(a/\mathbb{p}^{m})$ ] , which is smooth of dimension @xmath77 by theorem [ app ]
let @xmath85}(a/\mathbb{p}^{m})\to \mathbb{p}^{m}$ ] be the natural morphism , and set @xmath1088p_1+\cdots + [ \frac{h_r}{2}]p_r$ ] .
then the tangent space of the fiber of @xmath89 at @xmath87 $ ] has dimension @xmath62 by theorem [ genus_hilbert ] ( 2 ) , hence @xmath1089 .
k. cho , y. miyaoka , and n. shepherd - barron , _ characterizations of projective space and applications to complex symplectic geometry _ ,
higher dimensional birational geometry , adv .
pure math .
35 , math .
japan , 2002 , pp . 189 .
hwang , _ geometry of minimal rational curves on fano manifolds _ , school on vanishing theorems and effective results in algebraic geometry ( trieste , 2000 ) , ictp lect .
notes , vol . 6 , abdus salam int . cent
. theoret .
trieste , 2001 , pp .
335393 .
s. kebekus and l. sol conde , _ existence of rational curves on algebraic varieties , minimal rational tangents , and applications _ , global aspects of complex geometry , springer , berlin , 2006 , pp . 359416 .
y. miyaoka , _ numerical characterizations of hyperquadrics _ , complex analysis in several variables - memorial conference of kiyoshi oka s centennial birthday , adv .
pure math .
42 , math .
japan , 2004 , pp . | in this paper we address fano manifolds of dimension @xmath0 with a locally unsplit dominating family of rational curves of anticanonical degree @xmath1 . we first observe that their picard number is at most @xmath2 , and then we provide a classification of all cases with maximal picard number .
we also give examples of locally unsplit dominating families of rational curves whose varieties of minimal tangents at a general point is singular . |
let @xmath8 be an orientable 3-dimensional handlebody of genus @xmath9 with @xmath10 marked points .
we regard its boundary @xmath11 as a compact connected orientable surface @xmath12 of genus @xmath9 with @xmath10 marked points .
we denote by @xmath13 the _ complexity _ of @xmath6 , a measure which coincides with the number of components of a maximal multi - disk in @xmath6 .
we also define the complexity @xmath14 of @xmath2 as @xmath15 .
let @xmath0 be a finite simplicial graph . through this paper ,
we denote by @xmath16 and @xmath17 the vertex set and the edge set of @xmath0 respectively .
the _ right - angled artin group _ on @xmath0 is defined by @xmath18=1 { \rm ~if~and~only~if}~\{v_{i } , v_{j } \}\in e(\gamma ) \right\rangle.\end{aligned}\ ] ] for two groups @xmath19 and @xmath20 , we write @xmath21 if there is an embedding from @xmath19 to @xmath20 , that is , an injective homomorphism from @xmath19 to @xmath20 . similarly , we write @xmath22 for two graphs @xmath0 and @xmath23 if @xmath23 is isomorphic to an induced subgraph of @xmath0 .
we denote by @xmath7 and @xmath4 the handlebody group of @xmath6 and the mapping class group of @xmath2 respectively .
right - angled artin groups were introduced by baudisch @xcite .
recently , these groups have attracted much interest from @xmath24-dimensional topology and geometric group theory through the work of haglund - wise @xcite,@xcite on special cube complexes . in particular , various mathematicians investigate subgroups of right - angled artin groups or right - angled artin subgroups of groups .
crisp - sageev - sapir @xcite studied surface subgroups of right - angled artin groups .
kim - koberda @xcite proved that for any tree @xmath25 , there exists a pure braid group @xmath26 such that @xmath27 is embedded in @xmath26 .
bridson @xcite proved that the isomorphism problem for the mapping class group of a surface whose genus is sufficiently large is unsolvable by using right - angled artin subgroup in mapping class groups .
see koberda @xcite for other researches about right - angled artin groups and their subgroups . on the other hand , the geometry of mapping class groups of surfaces is well understood .
a handlebody group @xmath7 of @xmath6 is a subgroup of the mapping class group @xmath4 of @xmath2 .
hamenstdt - hensel @xcite showed that @xmath7 is exponentially distorted in @xmath4 .
therefore , the geometric properties of handlebody groups may be different from those of mapping class groups .
furthermore , disk graphs are not quasi - isometric to curve graphs ( see masur - schleimer @xcite ) .
our motivation of this article is whether the following three propositions are true when we change the assumptions of mapping class groups and curve graphs to handlebody groups and disk graphs .
( ( * ? ? ?
* theorem 1.1 and proposition 7.16))[main_koberda12 ] if @xmath28 , then @xmath29 .
( ( * ? ? ?
* theorem 2))[main_kim - koberda13 ] let @xmath2 be an orientable surface with @xmath30 .
if @xmath29 , then @xmath28 .
( ( * ? ? ?
* theorem 3))[second_kim - koberda13 ] let @xmath2 be an orientable surface with @xmath31 .
then there exists a finite graph @xmath0 such that @xmath29 but @xmath32 . [ standard_embeding ] an embedding @xmath33 from @xmath3 to @xmath7
is _ standard _ if @xmath33 satisfies the following two conditions . * the map @xmath33 maps each vertex of @xmath0 to a multi - disk twist ; * for two distinct vertices @xmath34 and @xmath35 of @xmath0 , the support of @xmath36 is not contained in the support of @xmath37 .
we first prove the following three theorems .
[ main_thm ] if @xmath38 , then @xmath39 .
[ second_thm ] let @xmath6 be a handlebody with @xmath40 or @xmath41 .
if @xmath39 , then @xmath38 .
let @xmath6 be a handlebody with @xmath42 . if there exists a standard embedding @xmath43 , then @xmath38 .
[ third_thm ] for @xmath44 and @xmath45 , there exists a finite graph @xmath0 such that @xmath39 but @xmath46 . from theorem [ third_thm ] , it follows that the converse of theorem [ main_thm ] is generally not true .
further , kim - koberda proved that having @xmath47-thick stars forces the converse of proposition [ main_koberda12 ] .
( ( * ? ? ?
* theorem 5))[third_kim - koberda13 ] suppose @xmath2 is a surface with @xmath48 and @xmath0 is a finite graph with @xmath47-thick stars . if @xmath29 , then @xmath28 .
we also prove the following : [ fourth_thm ] suppose @xmath6 is a handlebody with @xmath49 and @xmath0 is a finite graph with @xmath47-thick stars . if there is a standard embedding @xmath43 , then @xmath38 .
note that our all theorems also hold when we change handlebody groups to pure handlebody groups .
we also note that we can not apply the argument of kim - koberda @xcite for handlebody groups of high complexity handlebodies .
the methods in this paper are worthless for high complexity handlebodies .
when is the converse of theorem [ main_thm ] true ?
in this paper , a _ graph _ is a one - dimensional simplicial complex . in particular , graphs have neither loops nor multi - edges . for @xmath50 , the _ subgraph _ of @xmath0 _ induced by _
@xmath51 is the subgraph @xmath23 of @xmath0 defined by @xmath52 and @xmath53 in this case , we also say @xmath23 is an _ induced subgraph _ or a _ full subgraph _ of @xmath0 .
a graph @xmath0 is @xmath23-_free _ if no induced subgraphs of @xmath0 are isomorphic to @xmath23 . in particular
, @xmath0 is _ triangle_-_free _ if no induced subgraphs of @xmath0 are triangles .
the _ link _ of @xmath35 in @xmath0 is the set of the vertices in @xmath0 which are adjacent to @xmath35 , and denoted as @xmath54 .
the _ star _ of @xmath35 is the union of @xmath54 and @xmath55 , and denoted as @xmath56 .
clique _ is a subset of the vertex set which spans a complete subgraph . by a link , a a star , or a clique , we often also mean the subgraphs induced by them . for a positive integer @xmath47
, we say @xmath0 has @xmath47-_thick stars _
if each vertex @xmath35 of @xmath0 is contained in two cliques @xmath57 on @xmath47 vertices of @xmath0 whose intersection is exactly @xmath35 .
equivalently , @xmath54 contains two disjoint copies of complete graphs on @xmath58 vertices of @xmath0 for each vertex @xmath35 .
a _ handlebody _ @xmath59 of genus @xmath9 is a compact orientable @xmath24-dimensional manifold constructed by attaching @xmath9 one - handles @xmath60 to a @xmath24-ball , where @xmath61 is a @xmath62-disk and @xmath63 is an interval .
the boundary @xmath64 of @xmath59 is a closed connected orientable surface @xmath65 of genus @xmath9 .
a handlebody @xmath8 of genus @xmath9 with @xmath10 marked points is a handlebody of genus @xmath9 , together with @xmath10 pairwise distinct points @xmath66 , @xmath67 , @xmath68 , @xmath69 on @xmath64 .
we regard the boundary @xmath11 of @xmath6 as a compact connected orientable surface @xmath12 of genus @xmath9 with @xmath10 marked points .
a disk @xmath70 is _ properly embedded _ in @xmath6 if its boundary @xmath71 is embedded in @xmath11 , and its interior is embedded in the interior of @xmath6 .
a properly embedded disk @xmath70 is _ essential _ if the simple closed curve @xmath71 is essential in @xmath11 , that is , @xmath71 does not bound a disk in @xmath11 or is not isotopic to a marked point on @xmath11 . by a _ disk _ in @xmath6 we mean a properly embedded essential disk @xmath72 .
a _ disk twist _
@xmath73 along a disk @xmath70 in @xmath6 is the homeomorphism defined by cutting @xmath6 along @xmath70 , twisting one of the sides by @xmath74 to the right , and gluing two sides of @xmath70 back to each other ( see figure [ fig_disk_twist ] ) . along a disk @xmath70 in @xmath6 . ]
a _ multi - disk _ in @xmath6 is the union of a finite collection of disjoint disks in @xmath6 .
the number of components of a multi - disk is at most @xmath75 .
a multi - disk is _ maximal _ if the number of its components is @xmath75 .
the _ handlebody group _ @xmath7 of @xmath6 is the group of orientation preserving homeomorphisms of @xmath6 , fixing the marked points setwise , up to ambient isotopy .
the _ mapping class group _ @xmath4 of @xmath2 is the group defined by changing the role of homeomorphisms of @xmath6 into homeomorphisms of @xmath2 in the definition of the handlebody group . the _ pure handlebody group _ @xmath76 of @xmath6 is the group of orientation preserving homeomorphisms of @xmath6 , fixing the marked points pointwise , up to ambient isotopy .
we note that it is not important to distinguish between handlebody groups and pure handlebody groups in our considerations , since our all theorems hold for both handlebody groups and pure handlebody groups .
we call elements of @xmath7 or @xmath4 mapping classes .
an element @xmath77 of @xmath7 is a _ multi - disk twist _ if @xmath77 can be represented by a composition of powers of disk twists along disjoint pairwise - non - isotopic disks . .
] an element @xmath77 of @xmath4 is _ pseudo_-_anosov _ if @xmath78 for any isotopy class @xmath79 of simple closed curve on @xmath2 and @xmath80 ( see @xcite ) .
an element @xmath77 of @xmath7 is _ pseudo_-_anosov _ if its restriction @xmath81 to @xmath11 is a pseudo - anosov element of @xmath82 .
the _ support _ @xmath83 of a homeomorphism @xmath84 of @xmath6 is defined by @xmath85 similarly , we define the support of a homeomorphism @xmath84 of @xmath2 as @xmath86 the _ disk graph _
@xmath5 of @xmath6 is a graph whose vertex set is the set of isotopy classes of disks in @xmath6 .
two vertices are adjacent if the corresponding isotopy classes admit disjoint representatives .
the _ curve graph _
@xmath1 of @xmath2 is a graph defined by changing the role of disks in the definition of the disk graph into properly embedded essential simple closed curves in @xmath2 . by a _ curve _ in @xmath2
we mean a properly embedded essential simple closed curve in @xmath2 .
there exists a natural inclusion @xmath87 given by sending an isotopy class of a disk @xmath70 to the isotopy class of a curve @xmath71 . slightly abusing the notation
, we often realize isotopy classes of disks or curves as disks or curves .
to prove theorem [ main_thm ] , it is sufficient to show the following lemma .
[ lemma_of_main_thm ] let @xmath0 be a finite graph and @xmath6 a handlebody .
let @xmath88 be an embedding from @xmath0 to @xmath5 as an induced subgraph .
then for all sufficiently large @xmath47 , the map @xmath89 given by sending @xmath35 to the @xmath47th power @xmath90 of a disk twist @xmath91 along @xmath92 is injective .
we use the following lemma to prove lemma [ lemma_of_main_thm ] .
( ( * ? ? ?
* theorem 7 ( 1)))[lemma_of_main_thm_kim - koberda13 ] let @xmath0 be a finite graph and @xmath2 an orientable surface .
let @xmath93 be an embedding from @xmath0 to @xmath1 as an induced subgraph .
then for all sufficiently large @xmath47 , the map @xmath94 given by sending @xmath35 to the @xmath47th power @xmath95 of a dehn twist @xmath96 along @xmath97 in @xmath2 is injective .
for the proof , see @xcite .
let @xmath98 be an embedding as an induced subgraph . recall that @xmath5 is a subgraph of @xmath1 , and
so there is a natural embedding @xmath99 given by sending a disk @xmath70 to the boundary circle @xmath71 .
we set @xmath100 .
then @xmath93 is an embedding of @xmath0 into @xmath1 as an induced subgraph . by lemma [ lemma_of_main_thm_kim - koberda13 ] ,
there is a sufficiently large @xmath101 such that the map @xmath102 given by sending @xmath103 to @xmath95 is injective . since @xmath104 , a dehn twist @xmath96 along @xmath97
is extended to a disk twist @xmath91 along @xmath92 in @xmath6 .
therefore , the map @xmath105 given by sending @xmath103 to @xmath106 is injective . from lemma [ lemma_of_main_thm ] , @xmath3 is a subgroup of @xmath7 , and we have finished a proof of theorem [ main_thm ] .
the idea of the proof of theorem [ second_thm ] comes from the proof of ( * ? ? ?
* theorem 3 ) by changing the assumptions of mapping class groups and curve graphs to handlebody groups and disk graphs .
however , for @xmath107 and @xmath108 we can not apply their argument and we prove it by another way .
first we remark the following . [ remark_of_standard_embedding ] in definition [ standard_embeding ]
, if supp(@xmath33 ) is a maximal clique in @xmath5 , then the condition ( ii ) implies that @xmath35 is an isolated vertex .
first we consider the case @xmath40 , that is , @xmath109 and @xmath110 , or @xmath111 and @xmath112 .
if @xmath111 and @xmath112 , then there exists one disk in @xmath113 .
thus , this case comes down to the case @xmath41 .
we may assume that @xmath109 and @xmath110 .
then the handlebody groups are trivial .
note that there is no essential disk in @xmath6 .
we assume that @xmath3 is a subgroup of @xmath7 .
then @xmath3 is also trivial .
therefore @xmath3 has no generator , and so @xmath0 has no vertex .
hence @xmath38 .
suppose that @xmath41 , that is , @xmath114 , @xmath107 , or @xmath108 .
first , we assume @xmath114 .
note that @xmath5 is an infinite union of isolated vertices .
@xmath7 is virtually free , since @xmath4 is virtually free and subgroups of virtually free groups are also virtually free .
we assume that @xmath3 is a subgroup of @xmath7 .
then @xmath3 is free because it is virtually free and torsion - free .
hence , @xmath3 has no relation , and so @xmath0 is a graph consists of finite isolated vertices .
therefore @xmath38 .
secondly , we assume @xmath107 , or @xmath108 .
we note that there is only one essential disk in @xmath6 ( see hamenstdt - hensel ( * ? ? ?
* section 2 ) ) .
we call the disk @xmath70 .
hence , @xmath5 is a graph consists of a single vertex . on the other hand , @xmath115 , where @xmath116 is the group generated by a disk twist @xmath73 along @xmath70 ( see hamenstdt - hensel ( * ? ? ?
* proposition 2.2 ) ) .
we assume that @xmath3 is a subgroup of @xmath7 .
then , @xmath3 is trivial or @xmath117 because any non - trivial subgroup of @xmath117 is isomorphic to @xmath117 .
if @xmath3 is trivial , then @xmath0 has no vertex .
hence @xmath38 .
if @xmath3 is isomorphic to @xmath117 , then @xmath0 is a graph consists of a single vertex
. therefore @xmath38 .
suppose that @xmath42 , that is , @xmath118 or @xmath119 .
we note that @xmath5 is triangle - free .
first , we claim that the conclusion of the theorem holds for @xmath0 if and only if it holds for each connected component of @xmath0 .
this is an easy consequence of the fact that @xmath5 has infinite diameter and that there exists a pseudo - anosov homeomorphism on @xmath6 .
so , we may assume that @xmath0 is connected , and so it has at least one edge . by the hypothesis
, there is a standard embedding @xmath33 from @xmath3 to @xmath7 .
each vertex @xmath35 of @xmath3 is mapped to a power of a single disk twist @xmath73 along @xmath70 by @xmath33 , since @xmath0 has no isolated vertex and @xmath5 is triangle - free ( see remark [ remark_of_standard_embedding ] ) .
hence we gain an embedding @xmath120 .
let @xmath121 and @xmath122 be the finite graphs shown in figure [ fig_gamma_0_and_gamma_1 ] .
we denote by @xmath123 the 4-cycle spanned by @xmath124 . and @xmath122 .
] let @xmath125 be the map defined by @xmath126 and @xmath127 for any @xmath128 .
for a graph @xmath0 , we will denote by @xmath129 the subgroup of @xmath3 generated by @xmath103 .
( ( * ? ? ?
* lemma 11))[first_lemma_of_third_thm_kim - koberda13 ] the map @xmath125 is injective . for the proof see the proof of lemma ( *
* lemma 11 ) . , @xmath130 , and @xmath131 .
] [ embedding_gamma_1_into_disk_graphs ] the graph @xmath122 is embedded into @xmath5 if and only if @xmath6 is a handlebody with @xmath132 , @xmath44 , or @xmath45 .
we introduce the intersection number between two disks @xmath133 and @xmath134 in @xmath6 as follows .
@xmath135 two disks @xmath133 and @xmath134 in @xmath6 are in _ minimal position _ if the intersection number is minimal in the isotopy classes of @xmath133 and @xmath134 .
note that two disks in @xmath6 intersect if and only if the boundary circles intersect in @xmath11 .
we also remark that two curves are in minimal position in @xmath2 if and only if they do not bound any bigons on @xmath2 .
we show @xmath122 is embedded into @xmath136 , @xmath137 , and @xmath138 as an induced subgraph .
note that the complexities of @xmath139 , @xmath130 , and @xmath131 are four , five , and six respectively .
we put disks @xmath140 , @xmath141 , @xmath142 , @xmath70 , @xmath143 , @xmath33 , @xmath9 , and @xmath144 in the handlebodies as in figure [ fig_three_handlebodies ] so that they form the graph @xmath122 in the disk graphs .
one can verify that the disks are in minimal position , since their boundary circles do not bound bigons on the boundary surfaces .
hence @xmath122 is embedded into @xmath136 , @xmath137 , and @xmath138 as an induced subgraph .
therefore @xmath122 is also embedded into a disk graph of a handlebody whose complexity is at least six .
we can also show that @xmath122 is not embedded into any other disk graphs ( see section [ appendix ] for the proof ) .
we have thus proved the lemma .
let @xmath6 be a handlebody with @xmath145 or @xmath146 .
suppose \{@xmath140 , @xmath141 , @xmath142 , @xmath70 } are disks in @xmath6 which form a four cycle @xmath123 in @xmath5 with this order .
let @xmath147 be a regular neighborhood of @xmath148 and @xmath149 in @xmath11 , and @xmath150 a regular neighborhood of @xmath151 and @xmath71 in @xmath11 so that @xmath152 .
set @xmath153 .
note that we regard the boundaries of @xmath154 , @xmath147 , and @xmath150 as marked points from now .
[ 17_cases _ of_complexity_five_handlebodies ] let @xmath6 be a handlebody with @xmath146 .
then , the triple @xmath155 satisfies exactly one of the following seventeen cases , possibly after switching the roles of @xmath147 and @xmath150 .
* @xmath156 , @xmath157 , and @xmath154 intersects both @xmath147 and @xmath150 .
* @xmath158 , @xmath159 , and @xmath154 intersects each of @xmath147 and @xmath150 at only one boundary component .
* @xmath160 , @xmath161 , @xmath162 , and each component of @xmath154 intersects both @xmath147 and @xmath150 . *
@xmath160 , @xmath161 , @xmath162 , and one component of @xmath154 intersects each of @xmath147 and @xmath150 at only one boundary component , while the other component of @xmath154 intersects @xmath147 at just two boundary components .
* @xmath160 , @xmath161 , @xmath163 such that the @xmath164 component intersects both @xmath147 and @xmath150 and the @xmath165 component is disjoint from @xmath150 , and moreover , @xmath166 .
* @xmath167 , @xmath168 , and @xmath154 intersects each of @xmath147 and @xmath150 at only one boundary component .
* @xmath167 , @xmath169 such that the @xmath164 component intersects both @xmath147 and @xmath150 and the @xmath170 ( resp .
@xmath171 ) component is disjoint from @xmath150 , and moreover , @xmath172 ( resp .
@xmath173 ) .
* @xmath167 , @xmath159 , and @xmath174 and @xmath175 .
* @xmath167 , @xmath163 , and both components of @xmath154 intersects each of @xmath147 and @xmath150 at only one boundary component respectively .
* @xmath167 , @xmath163 such that the @xmath164 component intersects both @xmath147 and @xmath150 and the @xmath165 component is disjoint from @xmath150 , and moreover , @xmath166 .
* @xmath167 , @xmath163 such that the @xmath165 component intersects both @xmath147 and @xmath150 and the @xmath164 component is disjoint from @xmath150 , and moreover , @xmath176 .
* @xmath167 , @xmath177 , and three components of @xmath154 intersect both @xmath147 and @xmath150 . *
@xmath167 , @xmath177 , and two components of @xmath154 intersect both @xmath147 and @xmath150 , while the other component of @xmath154 is disjoint from @xmath150 and intersects @xmath147 at just two boundary components .
* @xmath167 , @xmath177 such that one component of @xmath154 intersects both @xmath147 and @xmath150 , one ( named @xmath63 ) of the other components is disjoint from @xmath150 , the other component ( named @xmath178 ) is disjoint from @xmath147 , and @xmath179 and @xmath180 . * @xmath167 , @xmath181 such that two @xmath164 components intersect both @xmath147 and @xmath150 and the @xmath165 component is disjoint from @xmath150 , and moreover , @xmath166 .
* @xmath167 , @xmath181 , such that one @xmath164 component intersects both @xmath147 and @xmath150 , the other @xmath164 component ( named @xmath63 ) is disjoint from @xmath150 , the @xmath165 component is disjoint from @xmath147 or @xmath150 ( here we suppose that @xmath165 is disjoint from @xmath150 ) , and moreover , @xmath179 and @xmath166 . * @xmath167 , @xmath182 such that the @xmath164 component intersects both @xmath147 and @xmath150 and each @xmath165 component is disjoint from @xmath147 or @xmath150 respectively ( here we suppose that both @xmath165 components are disjoint from @xmath150 ) , and moreover , each @xmath165 component intersects @xmath147 at only one boundary component . let @xmath79 be the number of free isotopy classes of boundary components of @xmath154 that are contained in @xmath183 .
we have @xmath184 , since @xmath2 is connected and @xmath152 .
let @xmath185 be the sum of complexities of the components of @xmath154 .
then @xmath186 .
since @xmath187 ( @xmath188 ) contains at least one curve , we have @xmath189 .
further , @xmath190 .
therefore it follows that @xmath191 .
we note that if @xmath147 or @xmath150 is a surface whose complexity is one , then it is homeomorphic to only @xmath170 .
in fact , if it is homeomorphic to @xmath171 , then it can not have two curves @xmath148 and @xmath149 since @xmath192 has only one isotopy class of disk .
we suppose that @xmath193 .
then we have @xmath194 . if @xmath195 and @xmath196 , then @xmath160 and @xmath197 .
if @xmath198 and @xmath199 , then @xmath200 and @xmath201 . by the assumption that @xmath202 , we have @xmath203 and @xmath204 . since @xmath154 has at least two boundary components , @xmath205 or @xmath159 . if @xmath159 , then this contradicts the assumption that @xmath203 .
hence , we have @xmath205 . case ( 1 )
is immediate .
we suppose that @xmath206 . then we have @xmath207 . without loss of generality
we may assume @xmath195 and @xmath199 .
it follows that @xmath160 and @xmath208 . if @xmath203 , then @xmath154 forced to be an annulus and we have a contradiction of the fact that @xmath207 .
so we have @xmath209 and @xmath204 .
if @xmath154 is connected , then @xmath209 implies that @xmath159 , and hence case ( 2 ) follows .
we assume that @xmath154 is not connected . by the assumption that @xmath209 , the number of connected components of @xmath154 is at most two . we have @xmath162 or @xmath163 . if @xmath162 , then we have two cases where each component intersects both @xmath147 and @xmath150 , and one ( we name it @xmath63 ) of the component is disjoint from @xmath150 . in the former case ,
we obtain case ( 3 ) . in the latter case , if @xmath210 , then there is no essential disk in @xmath63 , and so we have a contradiction to @xmath209 .
hence case ( 4 ) follows . if @xmath163 , then @xmath165 has to be disjoint from @xmath150 and @xmath166 since @xmath209
. then we obtain case ( 5 ) .
we suppose that @xmath211 .
then @xmath195 and @xmath212 .
thus @xmath160 and @xmath213 .
moreover it follows that @xmath214 .
if @xmath203 , then by a similar argument to that of case ( 1 ) we have @xmath205 , and this contradicts the fact that @xmath215 .
first , we consider the case where @xmath209 and @xmath216 .
if @xmath154 is connected , then @xmath209 implies that @xmath168 and @xmath172 , @xmath217 , hence case ( 6 ) follows .
we assume that @xmath154 is not connected . by the assumption that @xmath209 , the number of connected components of @xmath154 is at most two and the component which intersect both @xmath147 and @xmath150 has to be @xmath164 .
the other component ( we name @xmath63 ) which is disjoint from @xmath150 is @xmath170 or @xmath171 , and @xmath218 .
then case ( 7 ) follows .
next , we consider the case where @xmath219 and @xmath204 . if @xmath154 is connected , then @xmath219 implies that @xmath159 , @xmath220 , and @xmath221 .
hence case ( 8) follows ( see figure [ fig_case_8_9_10_11 ] ) .
we assume that @xmath154 is not connected . by the assumption that @xmath219 , the number of connected components of @xmath154 is at most three .
first we suppose that the number of connected components of @xmath154 is two . by the assumption that @xmath204 , @xmath162 or @xmath163 . if @xmath162 , then this contradicts the assumption that @xmath219 , and so we have @xmath163 . if each component of @xmath154 intersects both @xmath147 and @xmath150 , then case ( 9 ) is immediate ( see figure [ fig_case_8_9_10_11 ] ) . if the @xmath164 component intersects both @xmath147 and @xmath150 , and @xmath165 component is disjoint from @xmath150 , then case ( 10 ) is immediate ( see figure [ fig_case_8_9_10_11 ] ) . if the @xmath165 component intersects both @xmath147 and @xmath150 , and @xmath164 component is disjoint from @xmath150 , then case ( 11 ) is immediate ( see figure [ fig_case_8_9_10_11 ] ) .
finally , we suppose that the number of connected components of @xmath154 is three . by the assumption that @xmath204 , @xmath177 , @xmath181 , or @xmath182 .
we note that if @xmath222 , then this contradicts the assumption that @xmath219 .
we assume that @xmath177 .
if each component of @xmath154 intersects both @xmath147 and @xmath150 , then case ( 12 ) is immediate ( see figure [ fig_case_12_13_14_15 ] ) .
if two components of @xmath154 intersect both @xmath147 and @xmath150 and the other component ( we name it @xmath63 ) is disjoint from @xmath150 , then @xmath179 .
we obtain case ( 13 ) ( see figure [ fig_case_12_13_14_15 ] ) .
we assume that just one component of @xmath154 intersects both @xmath147 and @xmath150 .
we also assume that one of the other component ( we name it @xmath63 ) is disjoint from @xmath150 and the other component ( we name it @xmath178 ) is disjoint from @xmath147
. then @xmath179 and @xmath180 , and so case ( 14 ) follows ( see figure [ fig_case_12_13_14_15 ] ) .
we assume that @xmath181 .
note that the @xmath165 component of @xmath154 must not intersect both @xmath147 and @xmath150 since @xmath219 .
if each @xmath164 component intersects both @xmath147 and @xmath150 , then the @xmath165 component is disjoint from @xmath150 and @xmath166 , and case ( 15 ) is immediate ( see figure [ fig_case_12_13_14_15 ] ) .
we assume that just one of the @xmath164 component intersects both @xmath147 and @xmath150 , and the other component ( we name it @xmath63 ) is disjoint from @xmath150 .
we also suppose that the @xmath165 component is disjoint from @xmath150 . then @xmath179 and @xmath166 , hence case ( 16 ) follows ( see figure [ fig_case_16_17 ] ) .
we assume that @xmath182 .
note that each of the @xmath165 components of @xmath154 must not intersect both @xmath147 and @xmath150 since @xmath219 .
then the @xmath164 component intersects both @xmath147 and @xmath150 and two @xmath165 components are disjoint from @xmath150 , and moreover each @xmath165 component intersects @xmath147 at only one boundary component respectively .
we obtain case ( 17 ) ( see figure [ fig_case_16_17 ] ) .
[ five_cases_of complexity_four_handlebodies ] let @xmath6 be a handlebody with @xmath145 . then , the triple @xmath155 satisfies exactly one of the following five cases , possibly after switching the roles of @xmath147 and @xmath150 .
* @xmath223 , @xmath213 , @xmath157 , and @xmath154 intersects both @xmath147 and @xmath150 . *
@xmath167 , @xmath159 , and @xmath154 intersects each of @xmath147 and @xmath150 at only one boundary component .
* @xmath167 , @xmath162 , and each component of @xmath154 intersects both @xmath147 and @xmath150 . *
@xmath167 , @xmath162 , and one component of @xmath154 intersects each of @xmath147 and @xmath150 at only one boundary component , while the other component of @xmath154 intersects @xmath147 at two boundary components .
* @xmath167 , @xmath163 such that the @xmath164 component intersects both @xmath147 and @xmath150 and the @xmath165 component is disjoint from @xmath150 , and moreover , @xmath166 .
note that we obtain lemma [ five_cases_of complexity_four_handlebodies ] by changing the assumption of a surface @xmath2 in ( * ? ? ?
* lemma 13 ) to a handlebody @xmath6 .
we can prove this by the same process as the proof of lemma [ 17_cases _ of_complexity_five_handlebodies ] . in our case , if @xmath147 and @xmath150 are surfaces whose complexities are one , then they are homeomorphic to only @xmath170 . on the other hand , in (
* lemma 13 ) if @xmath147 and @xmath150 are surfaces whose complexities are one , then they are homeomorphic to @xmath170 or @xmath171 .
[ second_lemma_of_third_theorem ] for @xmath45 , there exists an embedding from @xmath224 to @xmath7 , but @xmath121 does not embed into @xmath5 as an induced subgraph .
first , we show the first half of the conclusion . by lemma [ first_lemma_of_third_thm_kim - koberda13 ] , @xmath225 . by lemma [ embedding_gamma_1_into_disk_graphs ] and lemma [ lemma_of_main_thm ] , @xmath226 .
therefore , it follows that @xmath227 . for the second half
, we suppose that @xmath228 .
we regard @xmath140 , @xmath141 , @xmath142 , @xmath70 , @xmath9 , @xmath144 , and @xmath229 as disks in @xmath6 . from @xmath230
, we have one of the seventeen cases in lemma [ 17_cases _ of_complexity_five_handlebodies ] . by the definition of disk graphs , the intersections @xmath231 , @xmath232 , @xmath233 , @xmath234 , and
@xmath235 are not empty . moreover @xmath236 and @xmath237 . in case
( 1 ) , the annulus @xmath154 connects @xmath147 and @xmath150 .
this implies that @xmath238 and @xmath239 , and so @xmath240 .
this is a contradiction . in case
( 2 ) , since @xmath241 , we have @xmath242 or @xmath243 . without loss of generality , we may assume @xmath242 .
the curve @xmath244 is a separating curve which separates @xmath147 from @xmath11 . by the fact that @xmath245
, we have @xmath246 .
this contradicts the fact that @xmath247 . in cases ( 3 ) , ( 4 ) , and ( 5 ) , if @xmath248 intersects @xmath249 , then they intersect on the components of @xmath154 which connect @xmath147 and @xmath150 . similarly to case ( 1 ) , we have @xmath240 and this is a contradiction . case ( 6 )
does not appear for @xmath130 .
note that we see @xmath121 is embedded in @xmath250 . in case ( 7 ) , by a similar argument to that of case ( 1 ) we have @xmath240 , and this is a contradiction . in case ( 8) , we have @xmath251 or @xmath243 since @xmath241 . without loss of generality
we may assume that @xmath251 .
by the fact that @xmath247 , @xmath248 intersects @xmath244
. then it follows that @xmath248 intersects @xmath147 .
this is a contradiction . in case
( 9 ) , by a similar argument to that of case ( 8) , we have @xmath247 .
this is a contradiction . in case
( 11 ) , by a similar argument to that of case ( 2 ) , we have @xmath246 .
this contradicts the fact that @xmath247 . in cases ( 10 ) , ( 12 ) , ( 13 ) , ( 14 ) , ( 15 ) , ( 16 ) , and ( 17 ) , by a similar argument to that of case ( 3 ) , we have @xmath240 .
this is a contradiction .
therefore it follows that @xmath252 .
( ( * ? ? ?
* lemma 14))[second_lemma_of_third_theorem_kim - koberda13 ] let @xmath2 be a surface with @xmath253 .
there exists an embedding from @xmath224 to @xmath4 , but @xmath121 does not embed into @xmath1 as an induced subgraph . for the proof see the proof of ( * ? ? ?
* lemma 14 ) .
[ first_lemma_of_third_theorem ] for @xmath44 , there exists an embedding from @xmath224 to @xmath7 , but @xmath121 does not embed into @xmath5 as an induced subgraph .
it directly follows from lemma [ second_lemma_of_third_theorem_kim - koberda13 ] , since the handlebody group of @xmath6 is isomorphic to the mapping class group of @xmath11 , and the disk graph of @xmath6 is isomorphic to the curve graph of @xmath11 . from lemmas [ second_lemma_of_third_theorem ] and [ first_lemma_of_third_theorem ] ,
we have completed the proof of theorem [ third_thm ] .
therefore the converse of theorem [ main_thm ] is not true in general .
[ notes for high complexity cases of handlebody groups ] for a graph @xmath0 , we define @xmath254 to be the minimum of @xmath255 among connected handlebodies @xmath6 satisfying @xmath38 . from lemma [ second_lemma_of_third_theorem_kim - koberda13 ] , we see @xmath256 .
the graph @xmath121 embeds into @xmath250 ( see figure [ fig_complexity_five_handlebody_embedded_gamma0_v2 ] , the handlebody of case ( 6 ) in lemma [ 17_cases _ of_complexity_five_handlebodies ] ) .
hence we see @xmath257 .
further , there exists a handle body @xmath6 whose complexity is four such that @xmath122 can not embed into @xmath5 . from these facts
, we can not apply the arguments of kim - koberda ( * ? ? ?
* section 4.3 ) for high complexity cases to handlebodies .
however , if @xmath6 is a handlebody whose complexity is @xmath258 and @xmath259 , then there exists a graph @xmath260 such that @xmath261 but @xmath262 by the same argument as that of the proof of ( * ? ? ?
* proposition 16 ) . is embedded into @xmath250 . ]
if @xmath263 is a multi - disk on @xmath6 , then we denote by @xmath264 a subgroup of @xmath7 which is generated by disk twists along the disks in @xmath263 .
the proof of theorem [ fourth_thm ] comes from the proof of kim - koberda ( * ? ? ?
* theorem 5 ) .
let @xmath35 be an arbitrary vertex of @xmath265 .
we write @xmath266 and @xmath267 for two disjoint cliques of @xmath0 such that @xmath268 and @xmath269 are cliques on @xmath47 vertices of @xmath0 .
we note that there are such cliques in @xmath0 for any @xmath103 because of the assumption .
the support of @xmath270 is a regular neighborhood of a multi - disk in @xmath6 , and we call the multi - disk @xmath263 . similarly , we write @xmath271 and @xmath272 for multi - disks in the supports of @xmath273 and @xmath274 in @xmath6 respectively . since @xmath49 , multi - disks @xmath275 and @xmath276 are maximal .
note that @xmath277 is a subgroup of @xmath278 . by the diagram in figure [ fig_diagram_v2 ]
, @xmath274 is a finite index subgroup of @xmath277 . by the fact that @xmath279 and @xmath280
, it follows that @xmath281 .
hence , for each @xmath103 , @xmath37 is some single disk twist @xmath73 along a disk @xmath70 , and we obtain an injection from @xmath0 into @xmath5 .
in this section , we prove the following proposition . first @xmath122
is not embedded into the disk graphs of @xmath6 with @xmath282 because @xmath122 has cliques on four vertices .
next we suppose that @xmath6 is a handlebody with @xmath145 , that is , @xmath44 , @xmath283 , or @xmath284 .
we will show that @xmath122 is embedded into only @xmath136 .
we suppose that @xmath286 . by the fact that @xmath287 , we obtain one of the five cases in lemma [ five_cases_of complexity_four_handlebodies ] . from the definition of disk graphs , we see @xmath288 , @xmath289 , @xmath290 , @xmath245 , and @xmath291 .
further , @xmath292 , @xmath293 , @xmath294 , @xmath295 , and @xmath296 . in case
( 1)@xmath297 , the annulus @xmath154 connect @xmath147 and @xmath150 .
this implies that @xmath298 and @xmath299 , and so @xmath240 .
this is a contradiction . in case ( 2)@xmath297 , @xmath147 and @xmath150
are homeomorphic to @xmath170 .
one can confirm that @xmath122 is embedded into only @xmath136 in this case .
note that if we discuss it for a surface @xmath2 with @xmath253 , then @xmath147 and @xmath150 are homeomorphic to @xmath170 or @xmath171 , and so @xmath122 is embedded into @xmath300 , @xmath301 , and @xmath302 . in cases ( 3)@xmath297 , ( 4)@xmath297 , and ( 5)@xmath297 , by the same argument as that of case ( 1)@xmath297 , we have @xmath240 , and this is a contradiction . secondly we suppose that @xmath6 is a handlebody with @xmath146 , that is , @xmath303 , @xmath45 , or @xmath285 . we will show that @xmath122 is embedded into only @xmath250 and @xmath137 . in cases ( 1 ) , ( 3 ) , ( 4 ) , ( 5 ) , ( 7 ) , ( 10 ) , ( 11 ) , ( 12 ) , ( 13 ) , ( 14 ) , ( 15 ) , ( 16 ) , and ( 17 ) , by the same argument that @xmath121 is not embedded into @xmath137 in the proof of lemma [ second_lemma_of_third_theorem ] , we have @xmath240 , and this is a contradiction . in case ( 2 ) , we see @xmath122 is embedded into only @xmath250 and @xmath137 . in case
( 6 ) , we see @xmath122 is embedded into only @xmath250 . in cases ( 8) and ( 9 ) ,
we see @xmath122 is embedded into only @xmath137 . by the argument above
, we have finished the proof of the proposition .
* acknowledgements : * the author is deeply grateful to hisaaki endo for his warm encouragement and helpful advice .
the author also wishes to thank susumu hirose and yeonhee jang for their meticulous comments and pieces of advice about handlebody groups and disk complexes . | koberda proved that if a graph @xmath0 is a full subgraph of a curve graph @xmath1 of an orientable surface @xmath2 , then the right - angled artin group @xmath3 on @xmath0 is a subgroup of the mapping class group @xmath4 of @xmath2 . on the other hand , for a sufficiently complicated surface @xmath2 , kim - koberda gave a graph @xmath0 which is not contained in @xmath1 , but @xmath3 is a subgroup of @xmath4 . in this paper
, we prove that if @xmath0 is a full subgraph of a disk graph @xmath5 of a handlebody @xmath6 , then @xmath3 is a subgroup of the handlebody group @xmath7 of @xmath6 .
further , we show that there is a graph @xmath0 which is not contained in some disk graphs , but @xmath3 is a subgroup of the corresponding handlebody groups . |
recently the cleo collaboration has reported measurements on the branching ratios of rare hadronic b decays @xmath7@xcite .
the data have attracted great interest from both theorists and experimentalists .
the study of these channels will provide us important insights on understanding the effects of electroweak penguins ( ewp ) in b - system@xcite , and the final state interactions(fsi)@xcite , as well as extracting the weak ckm phase @xmath8@xcite .
it may also open a window for probing new physics@xcite . from the current data
, it is noticed that the branching ratio for @xmath9 is relatively small , @xmath10 ( in units of @xmath11 ) .
the decays for @xmath12 have almost equal decay rates , i.e. , @xmath13 . while an unexpectedly large branching ratio for @xmath14 decay was also observed , @xmath15 .
these measurements seem in conflict with the calculations based on the naive factorization hypotheses . for the first three decays ,
it was pointed out that the factorization approach may still be valid if one takes the weak phase @xmath16 to be greater than @xmath17@xcite .
with such a large @xmath16 , i.e. , @xmath18 , the interference between tree and penguin diagrams has opposite sign in @xmath0 and @xmath19 decays .
thus the negative @xmath20 will suppress the decay rate for @xmath21 and enhance that for @xmath22 . as a consequence ,
the almost equal decay rates for @xmath23 and @xmath24 decay modes indicate the dominance of strong penguin . however , the large rate for @xmath25 is not easily explained .
most recent analysis showed that a large final state interaction ( fsi ) phase would be helpful to enhance the branching ratio for the @xmath25 decay@xcite , but only considering the elastic - rescatterings remains insufficient to obtain the large central value of the data . to understand the measured data , besides some model - dependent calculations , a model - independent approach using a single relative strong phase has also been proposed for the study of @xmath26 decays@xcite . since the ordinary factorization approach suffers from the uncertainties due to hadronic matrix elements , such as the meson decay constants , b - meson form factors and so - called effective color number @xmath27 .
the model - independent analyses may be more useful as more data become available .
the approach based on the isospin @xmath28 and approximate flavor @xmath29@xcite symmetries of the strong interactions has been proposed to constrain @xcite and extract the weak phase @xmath16@xcite .
it has been noticed that the ratios between the cp - averaged decay rates , such as @xmath30 may provide us important information on the weak phase @xmath16 .
most recently , it has been shown that the weak phase @xmath16 may be determined through three ratios among cp - averaged decay rates of @xmath31 , @xmath32 , @xmath33 and @xmath34 decays@xcite . where two solutions were obtained at the @xmath5 level , one with positive @xmath35 and negative @xmath20 , i.e. , relative small strong phase @xmath36 and large weak phase @xmath16 , and another with negative @xmath35 and positive @xmath20 .
the latter with positive @xmath20 seems to be favored by solutions obtained from other constraints in the standard model but appears not to be as favorable as the one with negative @xmath20 studies of all the existing charmless decays are taken in account .
however , there is still no complete analysis in the literature . in this paper
, we shall give a general analysis for all 7 decay modes of @xmath0 and @xmath1 .
for that purpose , we will start from a general model independent parameterization for all the decay amplitudes by considering both the isospin and simple diagrammatic decompositions .
we will show that there are in general 15 independent variables . by assuming the su(3 )
relations which appropriately account for su(3 ) symmetry breaking effects , it allows us to reduce the 15 independent variables into 9 .
they consist of 6 isospin amplitudes and 2 relative strong phase as well as one weak phase @xmath16 .
note that once the relative strong phase is zero , @xmath37 only receives contributions from penguin - type diagram and the amplitudes of the isospin @xmath38 and @xmath39 amplitudes from tree - type diagrams cancel each other , which provides an additional constraint@xcite .
in fact , one of the isospin amplitudes becomes almost irrelevant due to the suppression factor of the ckm mixing element .
with such a consideration , there are only 8 relevant unknown quantities with 5 isospin amplitudes and 3 phases .
we show that the current 6 measured decay rates allow us to extract 6 unknown quantities as functions of two variable .
the upper bound of the decay rate @xmath40 also provides a bound for the difference between the two strong phases . once taking the numerical value of the weak phase @xmath16 to be the one obtained from other constraints in the standard model and fixing one of the strong phases ,
all the other parameters can be determined . with these determined parameters , we are then able to predict the branching ratio of the @xmath6 decay mode which is yet unmeasured due to the difficulty of its identification by the current detector .
in addition we also present predictions for direct cp violations in all 7 decay channels of @xmath41 . in our numerical fitting ,
we have adopted the @xmath2 analysis for the cleo data in order to have a systematic treatment on the experimental errors . in general , according to the watson theorem , there are two independent relative strong phases associating with the isospin amplitudes .
they are often assumed to be equal in the literatures@xcite . in this work ,
we shall make a more general analysis with two relative strong phases .
it is shown that the equal phase assumption will result in large enhancement of isospin amplitude @xmath42 which will be 5 times larger than the one calculated from the factorization approach .
the value of the strong phase is found to be @xmath43 .
these large values may imply large inelastic fsi or indicate the possible new physics effects
. however , if the two strong phases are different , the value of @xmath42 can be lower and is comparable with the usual factorization calculations .
it is remarkable to observe that within @xmath5 all the 6 decay rates can be consistently fitted for a large range of the weak phase @xmath44 for the above two cases .
it is also of interest to note that one of isospin amplitudes and the strong phases have a weak dependence on the weak phase @xmath16 .
three isospin amplitudes show a moderate dependence on the weak phase @xmath16 .
only one isospin amplitude is sensitive to @xmath16 . in particular ,
the fitting values for the 4 usual isospin amplitudes considered in most of the literature could still be comparable with the ones obtained by using naive factorization approach .
the resulting large strong phases may be regarded as a strong indication of large fsi in @xmath7 decays .
we begin with writing the decay amplitude of @xmath26 in the following general form : a^(k)= ^d(s)_u a^(k)_u+^d(s)_c a^(k)_c , where @xmath45 and @xmath46 are the products of ckm matrix elements .
the term proportional to @xmath47 has been absorbed into the above two terms by using the unitarity relation , @xmath48 .
we also find it useful to adopt the isospin decomposition for the decay amplitudes a^u , c_^-^+&= & + [ su2 - 1 ] + a^u , c_^0 ^ 0&= & - + a^u , c_^-^0&=&- + a^u , c_^+k^-&= & + + a^u , c_^0|k^0&= & - + a^u , c_^0k^-&=&- - a^u , c_^-|k^0 [ su2 - 6 ] with @xmath49 where @xmath50 and @xmath51 are the isospin amplitudes and @xmath52 and @xmath53 are the strong phases due to final state interactions . in some literatures
the strong phase of isospin amplitude @xmath54 is assumed to be equal to the one of @xmath55 for simplicity@xcite .
however , in the most general case , these strong phases are not necessarily the same , since they arise from the effective hamiltonian with different isospin .
the subscript @xmath56 denote the isospins of the amplitudes .
the advantage of the isospin decomposition allows one to use @xmath29 relations including leading order @xmath29 breaking effects . in other words ,
the isospin amplitudes are assumed to satisfy the following relations : a^u , c_0&&(f_/f_k ) a^u , c_1/2 , a^u , c_2(f_/f_k ) a^u , c_3/2 + _ 0&&_1/2 , _ 2 _ 3/2 where @xmath57 and @xmath58 are the @xmath59 meson decay constants with @xmath60 . for convenience , we define two phase differences as follows : & = & _ 3/2-_1/2 + & = & _1/2-_1/2 practically , the decay amplitudes are evaluated by calculating various feynman diagrams . in order to see how those isospin amplitudes receive contributions from diagrams , we also present a simple diagrammatic decomposition .
the diagrams can in general be classified into 6 types denoted by @xmath61(tree ) , @xmath62(color suppressed tree ) @xmath63(qcd penguin ) , @xmath64(electroweak penguin ) and @xmath65 ( color suppressed electroweak penguin ) @xcite : a_^-^+ & = & t+p+23p^c_ew + a_^0 ^ 0 & = & ( -c+p - p_ew-13p^c_ew ) + a_^-^0 & = & ( -t - c - p_ew - p^c_ew ) + a_^+k^- & = & t+p+23p^c_ew + a_^0|k^0 & = & ( -c+p-p_ew - p^c_ew ) where the primed and unprimed quantities are the amplitudes in @xmath66 and @xmath67 decays .
they roughly differ by a factor @xmath60 when the @xmath29 flavor symmetry breaking effects are considered . combining the two decompositions
, it is straight forward to get the following relations : & = & 2t-c+3p-p_ew+p^c_ew^u , c , + & = & t+c+p_ew+p^c_ew ^u , c [ a2c ] . from the above equations ,
one may easily see the relative magnitudes among those @xmath28 invariant amplitudes .
if the inelastic rescattering effects are small , @xmath68 will only contribute to the term proportional to @xmath69 .
therefore one may expect that @xmath70 .
this will lead to @xmath71 .
since @xmath72 receives contributions from qcd penguins , while @xmath42 only gets contributions from ewps , one may conclude that @xmath73 . to obtain relations for the isospin amplitude @xmath54
, one needs to be careful in adopting the diagrammatic decomposition implied by the naive factorization ansatz .
this is because the resulting relative strong phase is zero in the factorization approach , i.e. , @xmath74 and @xmath75 . as a consequence , @xmath76
the amplitudes with isospin @xmath38 and @xmath39 from tree - type graphs cancel each other in eq.([penguin ] ) .
thus the total amplitude only receives contributions from penguin diagrams in this case , namely @xmath77 if assuming the @xmath78-quark dominance in the penguin loops , one finds from eq.([b1/2 ] ) that : a^u_^-|k^0 a^c_^-|k^0 or b^u_1/2b^c_1/2 + a^u_3/2 - a^c_3/2 .
[ approx ] which may be assumed for simplicity to be approximately valid after considering final state interactions with nonzero strong phases . in the numerical calculations ,
we have check that the amplitude @xmath79 is less important due to the strong suppression of the ckm factor ( for instance , even if taking @xmath80 , the results remain almost unchanged ) . with the above analyses ,
let us provide an intuitive discussion on how to yield a large branching ratio for @xmath14 decay by appropriately choosing the isospin amplitudes .
note the fact that @xmath81 , one may roughly estimate the ratio between @xmath82 and @xmath83 by neglecting the terms containing the ckm factor @xmath69 r= & & ^2 , [ r ] with @xmath84 .
once neglecting @xmath42 , the ratio may be simply given by @xmath85 , which is much smaller than the central value of the data @xmath86 .
it indicates that to enhance the decay rate of @xmath87 , the isospin amplitude @xmath42 should not be neglected .
its small value may provide a sizable contribution for a large value of @xmath88 .
this is because in this case there exists a constructive interference between @xmath72 and @xmath42 in @xmath87 and a destructive interference in @xmath23 .
the situation is quite similar to the case for a large @xmath89 , which is considered to enhance @xmath90 and decrease @xmath0 decay rates . from eq.([r ] ) , it is easily seen that the value of @xmath42 satisfies a^c_3/2 a^c_1/2 0.12 a^c_1/2 . with the above considerations , there are only 8 relevant quantities : @xmath91 , which should be constrained by six measured decay rates and one upper bound .
when taking the weak phase @xmath16 as a free parameter , the rest 6 variables can be determined from 6 equations of eqs.([su2 - 1]-[su2 - 6 ] ) .
as the errors in the current data remain considerable large , one may not take the central values of the data to be too serious .
thus by only using the central values of the data to determine the 6 variables may not be good enough . to take into account the experimental errors in a systematic way
, we shall adopt a global @xmath2 ( least square ) analysis for the present data .
this treatment allows us to obtain not only the central value but also the errors for the fitting amplitudes .
our fitting will be carried out by using the standard @xmath2 analysis program package minuit@xcite .
in order to compare with the values estimated from the factorization , it is necessary to explicitly see how large of the isospin amplitudes @xmath92 and @xmath93 from the factorization calculations , we present the relevant formulae for the @xmath7 decay amplitudes with the assumption of factorization@xcite : a_^+k^-&=&^s_u a_1-^s_t(a_4+a_10+(a_6+a_8)r_4 ) + a_^0|k^0&=&-^s_t a_4+a_6r_5-(a_10+a_8r_5 ) + & & -^s_u a_2-^s_t ( a_9-a_7 ) + a_^0k^-&=&-^s_u a_1- ^s_t(a_4+a_6r_4+(a_10+a_8r_4 ) + & & -^s_u a_2-^s_t ( a_9-a_7 ) + a_^-|k^0^c & = & -^s_t a_4+a_6r_5-(a_10+a_8r_5 ) where @xmath94 and @xmath95 are the decay constants and @xmath96-meson form factors respectively , @xmath97 and @xmath98 . in the flavor @xmath28 limit , one has @xmath99 .
the expressions of the isospin amplitudes can be rewritten as follows : a^u_1/2e^i_1/2&=&a_^+k^-^u + a_^0|k^0^u + & = & r ( a_1+a_4+a_10+(a_6+a_8)r_4 ) + ( a_4+a_6 r_5 -(a_10+a_8r_5 ) ) .
+ & & . - ( a_2+(a_9-a_7)x ) + a^c_1/2e^i_1/2&=&a_^+k^-^c + a_^0|k^0^c + & = & r(a_4+a_10+(a_6+a_8)r_4 ) + ( a_4+a_6 r_5 -(a_10+a_8r_5 ) ) .
- ( a_9-a_7)x + a^u_3/2e^i_3/2&=&a_^+k^-^u - a_^0|k^0^u + & = & r a_1+a_2x+(a_9-a_7)x+(a_10+a_8)r_4 + a^c_3/2e^i_3/2&=&a_^+k^-^c - a_^0|k^0^c + & = & r ( a_10+a_8r_4+(a_9-a_7)x ) and b_1/2^c - a_3/2^c & = & a_^-|k^0^c + & = & r a_4+a_6r_5-(a_10+a_8r_5)where @xmath100 and @xmath101 . in our numerical estimates , we will take @xmath102 mev , @xmath103 mev , @xmath104 and @xmath105 there remains a large uncertainty in strange quark mass @xmath106 . for @xmath107 mev
, we find that the numerical values of those amplitudes are given by a_1/2^u & & 818~846 , a_3/2^u 709 , + a_1/2^c & & - ( 103~131 ) , a_3/2^c -7 , + b_1/2^c & & -(72 ~100 ) . the results from the @xmath2-fitting are shown in fig.[iso00.eps]-[iso60.eps ] , where the six amplitudes as well as their errors at 1@xmath108 level are obtained as functions of the weak phase @xmath16 with @xmath109 fixed at @xmath110 respectively .
the relative magnitudes of the amplitudes are consistent with the previous discussions . in our fit , the minimum value of @xmath2 is found to be extremly low ( typiclly @xmath111 ) .
this means that the @xmath2-fits are highly consistent and the 6 amplitudes are actually extracted as the solutions of eqs.2 - 8 .
it can be seen from the figures that the @xmath16 dependence of @xmath112 is quite strong and the one of @xmath113 and @xmath114 is relatively weak . on the other hand
, it may be used to extract the angle @xmath16 once one of those amplitudes can be determined or calculated in other independent ways .
it is of interest to see that the @xmath16 dependence of the amplitude @xmath115 and the strong phase @xmath36 is weak , which shows that these two quantities are approximately fixed .
the possibility of large @xmath36 was also suggested in ref.@xcite to explain the large branching ratio of the @xmath25 decay .
recently , perturbative qcd calculations have also shown a large strong phase@xcite . in fig.[iso00.eps ] where the phase difference @xmath109 is set to be zero as usual
, the @xmath2-fitting shows that the values of @xmath112 and @xmath116 may be comparable with the ones from the theoretical estimations .
only when the weak phase @xmath16 is large . especially for @xmath117 , the fitting values could coincide with the ones from factorization except for @xmath118 . for @xmath119 ,
the two amplitudes are smaller than the ones from naive factorization calculations .
it appears that the factorization approach may become suitable for large weak phase @xmath16 .
this phenomenon was observed by most of the analyses in the literature which neglects the isospin amplitude @xmath118 . as a consequence
, the resulting large value of @xmath16 seems to be conflict with the one obtained from other constraints in the standard model . before drawing the final conclusion ,
one also notice that in the factorization approach , one yields a zero strong phase @xmath120 which actually contradicts with the general fitting value @xmath43 .
therefore , the results of estimates based on the naive factorization approach should be unreliable , and the isospin amplitudes must receive additional large contributions .
a large value for the relative strong phase @xmath121 implies that the final state interactions or inelastic rescattering effects must be significant .
we would like to stress that the most outstanding feature of the @xmath2 analysis with @xmath75 is that the isospin amplitude @xmath42 is likely to be relatively larger than the one estimated from the naive factorization calculations .
the fitting central value of @xmath42 may be larger by a factor of @xmath122 . to explicitly see how the decay rates depend on the isospin amplitude @xmath42 , we plot in fig.[a2c.eps ] the 6 branching ratios of @xmath7 decays as functions of @xmath42 .
it can be clearly seen that if @xmath75 , a small value of @xmath123 is not able to reproduce all the cleo data within the @xmath5 level , especially the data for the channels of @xmath124 and @xmath125 . to consistently describe the whole data , we need a relative large value @xmath126 for fitting the central value of the data which is about @xmath127 of the largest one @xmath128 . within the standard model
, it seems difficult to enhance the isospin amplitude @xmath42 by an order of magnitude even the inelastic fsi is involved , this is because the main inelastic channels such as @xmath129 , only contribute to isospin-@xmath130 part of the decay amplitude . in the sm , it is known that the ratio @xmath131 can be determined without the hadronic uncertainties in the flavor @xmath29 limit , this is because the ratio only depends on the short distance wilson coefficients@xcite .
thus a large value of @xmath131 may indicate the existence of new physics . while all models beyond the standard model must effectively provide large contributions to the electroweak penguins in order to enhance the isospin amplitude @xmath42 , such models are : susy with r parity violation , @xmath132-model and @xmath133-mediated fcnc models , et .
let us now consider the case that @xmath109 is non - zero , the situation then becomes quite different . in fig.[iso30.eps ] and fig.[iso60.eps ] , it is seen that @xmath42 decreases as the value of @xmath109 increases .
when @xmath109 reaches @xmath134 , @xmath42 will be consistent with the value yielded from factorization approach .
it is also noticed from the figures that a large @xmath109 leads to large values of @xmath135 and @xmath36 .
the enhancement of @xmath135 may be easily understood than the enhancement of @xmath42 due to final state interactions . nevertheless , as will be discussed below
, the values of @xmath109 can not be too large due to the constraint of the upper bound of the branching ratio of @xmath136 .
when all the isospin amplitudes and strong phases are determined , one is able to predict the direct cp asymmetries for all the relevant decay channels .
the direct cp asymmetry in @xmath7 decays is defined in the standard way a_cp = a_^f where @xmath137 denotes the final state mesons . in fig.[acp.eps ] we plot several @xmath138 s as functions of the weak phase @xmath16 with different value of @xmath109 .
when @xmath16 is near @xmath139 and @xmath75 one has @xmath140 which is in a good agreement with the most recent cleo data @xmath141@xcite . at this point , we have a reliable prediction for the direct cp violations in the following decay modes .
the @xmath142 s with @xmath143 read @xmath144 for @xmath145 , and @xmath146 for @xmath147 . in spite of the large @xmath36 ,
the smallness of cp asymmetries in @xmath148 decay is due to the absence of the interference between tree and penguin diagrams .
there remains an unobserved decay mode in @xmath7 decays , which is the decay mode @xmath6 ( the cleo collaboration has already reported the indication of @xmath149 ) .
as all the relevant isospin amplitudes have been determined as functions of @xmath16 , it allows us to predict @xmath6 as a function of @xmath16 .
it is interesting to note that our @xmath2-analysis shows that the resulting branching ratio @xmath150 is almost independent of the weak phase @xmath16 .
its value at @xmath75 is close to the one of @xmath9 decay @xmath151 with @xmath109 increasing , the branching ratio becomes larger and can reach @xmath152 when @xmath153 .
such a large branching ratio is about an order of magnitude larger than the prediction based on factorization calculations .
most recently , the cleo collaboration reported an upper bound of @xmath154@xcite , this will impose a strong constraint on the value of @xmath109 ( see fig .
it is seen that to be consistent with the data at the @xmath5 level , the upper bound of the branching ratio @xmath150 limits @xmath109 to be less than @xmath155 .
in summary , we have made a general less model - dependent investigation on the charmless b - meson decays by using the @xmath2 analysis based on the most recent cleo data .
we have used the most general isospin decomposition with two independent strong phases .
all the isospin amplitudes in rare hadronic @xmath96 decays @xmath7 can be determined as functions of the weak phase @xmath16 and one strong phase @xmath109 .
the effects of two equal and unequal strong phases are studied in detail .
it is found that the isospin amplitude @xmath115 and the strong phase @xmath36 only slightly depend on the phase @xmath16 .
an important observation under the equal strong phase assumption is the relative large isospin amplitude @xmath42 , @xmath156 where the central value is about 5 times greater than the one obtained from the factorization calculations .
when the two strong phases are not equal , the allowed values of @xmath42 decrease as their difference , i.e. , @xmath109 , increases . for the most general case with two rather than one large fsi strong phases , the magnitude of all the isospin amplitudes may be around the one estimated from the factorization approach
nevertheless , one needs to find out the mechanism of producing large strong phases .
this could directly be tested by measuring the branching ratio @xmath150 .
the direct cp asymmetries @xmath157 for all the relevant decay channels have also been given as functions of @xmath16 .
the resulting numerical value for @xmath158 is consistent with the most recent data . when taking @xmath16 to be in a reasonable range @xmath159 , we are led to the results given in eqs.(35 ) and ( 36 ) , which can be directly tested by experiments in the near future .
a resulting large branching ratio @xmath150 which is comparable with the one @xmath160 will also provide an important and consistent test . from the most general analysis presented in this paper
, the data appears to strongly suggest that final state interactions and inelastic rescattering effects must be significant and play an important rule in the charmless @xmath7 decays .
otherwise , our general analyses may be interpreted as hinting at the existence of new physics . for a more definite conclusion
, one needs more precise data .
the two b - factories babar and belle are expected to provide us with more information from the charmless decays . the author ( c.q.g . ) was supported by the nsc under contract number nsc 89 - 2112-m-007 - 013 .
is partially supported by the natural science and engineering council of canada .
two of us ( y.l.w . and y.f.z . ) were supported in part by nsf of china under grant no .
@xmath161 19625514 .
y.l.w acknowledges the cts for the support during his visit and he would also like to thank h.y .
cheng , x.g .
hou and h.n .
li for useful discussions .
cleo collab .
hep - ex/0001010 .
cleo collab .
hep - ex/0001009 .
g. kramer , w.f.palmer .
+ n. g. deshpande and x. g. he , .
w. s. hou and k. c. yang hep - ph/9911528 , to appear in @xmath162 r. fleischer , .
m. gronau and j. l. rosner + m. gronau and j. l. rosner .
m. neubert and j. l. rosner .
n. g. deshpande , x. g. he , w. s. hou and s. pavaska ; + x. g. he , w. s. hou and k. c. yang ; + w. s. hou and k. c. yang hep - ph/9908202 .
t. n. pham , hep - ph/0009142 , + c. isola , t. n. pham , hep - ph/9911534 . y. l. wu and y. f. zhou , phys .
* d62 * 036007 ( 2000 ) .
x. g. he , c. l. hsueh and j. q. shi ; + y.l . wu and y.f .
zhou , phys . rev .
* d61 * 096001 ( 2000 ) .
d. zppenfeld , ; + m. savage and m. wise , ; + m. savage and m. wise , ; + m. gronau , o. f. hernandez , d. london and j. l. rosner , ; + m. gronau , o. f. hernandez , d. london and j. l. rosner , ; + x.g .
j. * c9 * ( 1999 ) 443 ; + m. gronau and j. l. rosner .
r. fleischer and t. mannel , .
a.j . buras and r. fleischer hep - ph/0003323 . m. gronau and j. l. rosner .
f.james and m.roos , @xmath163 * 10*,343(1975 ) + cernlib - writeup , http://wwwinfo.cern.ch/asdoc/shortwrupsdir/d506/top.html y.y .
keum , h .-
li and a.i .
sanda , talk given at 28th intern .
workshop on gross properties of nuclei and nuclear excitation : hadrons in dense matter ( hirschegg 2000 ) , hirschegg , austria , 16 - 22 jan 2000 , hep - ph/0004173 .
a. ali , g. kramer and c.d .
l@xmath164 , .
m.neubert , j.rosner , .
d.hitlin , et al , babar collaboration , plenary talk in ichep 2000 ( osaka , japen , july 21,2000 ) , slac - pub-8540 .
+ h.aihara , belle collaboration , plenary talk in ic hep 2000 ( osaka , japen , july 21,2000 ) , cleo collaboration , plenary talk in ichep 2000 ( osaka , japen , july 21,2000 ) . | -1 cm we present a general analysis on charmless b - meson decays @xmath0 and @xmath1 .
it is noticed that the final state interactions and inelastic rescattering effects must be significant in order to understand the consistency of the current data . by using general isospin decompositions , the isospin amplitudes and the corresponding strong phases
could be extracted from a global @xmath2 fit of the experimental data .
we emphasize that in general there are two , rather than one , relative strong phases in the decomposition . in the assumption of two equalstrong phases as considered in the literature , the current data , especially the ones concerning @xmath3decays , will imply a large isospin amplitude @xmath4 , which is larger by a factor of five than the one from the naive factorization estimation .
when two different strong phases are considered , all the isospin amplitudes can become , within the @xmath5-level , comparable with the theoretical values .
we also show that the difference between the two strong phases can not be too large and will be restricted by the most recent upper bound of @xmath6 decay . in any case , the strong phases are found to be large and the branching ratio of @xmath6 is likely to be enhanced by an order of magnitude in comparison with the one obtained from the naive factorization approach .
direct cp violations in all decay modes are also calculated and found to be close to the sensitivity of the present experiments .
# 1#2#3 _ nucl . phys . _ * b#1 * ( # 2 ) # 3 # 1#2#3 _ nucl .
phys .
proc .
suppl . _
* # 1 * ( # 2 ) # 3 # 1#2#3 _ phys .
lett . _
* b#1 * ( # 2 ) # 3 # 1#2#3 _ phys .
rev .
_ * d#1 * ( # 2 ) # 3 # 1#2#3 _ phys .
rep . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ phys .
rev .
lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ int .
j. mod . phys._*a#1 * ( # 2 ) # 3 # 1#2#3 _ mod .
phys .
lett . _ * a#1 * ( # 2 ) # 3 # 1#2#3 _ zeit .
fr physik _
* c#1 * ( # 2 ) # 3 # 1#2#3_phys .
rep_*#1*(#2 ) # 3 # 1#2#3_eur .
phys .
j._*c#1*(#2)#3 # 1#2#3_jhep._*c#1*(#2)#3 # 1| # 1 | # 1#2#1_#2^u , c e^i _ # 2 -1 cm -1 cm |
the main motivation for research and development efforts on muon collider technology is the assertion that affordably priced muon colliders might provide lepton - lepton collisions at much higher center of mass ( com ) energies than is feasible for electron colliders , and perhaps eventually explore the spectrum of elementary particles at mass scales inaccessible even to hadron colliders .
this paper attempts to present some justification for these assertions through discussion and evaluation of the self - consistent muon collider parameter sets given in table 1 , at com energies ranging from 0.1 to 100 tev .
the parameter set at 0.1 tev com energy was included as a lower energy reference point and was constrained to essentially reproduce one of the sets of parameters currently under study[1 ] by the muon collider collaboration ( mcc ) .
in contrast , the other parameter sets represent speculation by the author on how the parameters might evolve with com energy and they have not been studied or discussed in detail within the mcc .
the parameter sets in table 1 were generated through iterative runs of a stand - alone fortran program , lumcalc .
the parameter sets are calculated from the input values for several input parameters namely , the com energy ( @xmath0 ) , the collider ring circumference ( c ) and depth below the earth s surface ( d ) , the beam momentum spread ( @xmath1 ) and 6-dimensional invariant emittance ( @xmath2 ) , the reference pole - tip magnetic field for the final focus quadrupoles ( @xmath3 ) , and the time until the beams are dumped ( @xmath4 ) and from the input of maximum allowable values for several other parameters namely , the bunch repetition frequency ( @xmath5 ) , the initial number of muons per bunch ( @xmath6 ) , the beam - beam tune disruption parameter ( @xmath7 ) , the beam divergence at the interaction point ( @xmath8 ) , the maximum aperture for the final focus quadrupoles ( @xmath9 ) , and maximum allowable neutrino radiation where the plane of the collider ring cuts the earth s surface . as a preliminary stage of calculation
, lumcalc makes any parameter adjustments that may be required to satisfy the input constraints .
these are , in order : 1 ) reducing @xmath8 to the limit imposed by @xmath9 ( based on scaling to existing final focus designs at 0.1 tev and 4 tev[1 ] ) , 2 ) reducing @xmath10 to attain an acceptable @xmath7 , and 3 ) reducing @xmath5 until the neutrino radiation is acceptable .
the output luminosity may be derived in terms of the input parameters as : @xmath11 } & = & { \rm 2.11 \times 10^{33 } \times h_b \times ( 1-e^{-2t_d[\gamma \tau_\mu ] } ) } \nonumber \\ & & \times { \rm \frac { f_b[s^{-1 } ] ( n_0[10^{12}])^2 ( e_{com}[tev])^3 } { c[km ] } } \nonumber \\ & & { \rm \times \left ( \frac{\sigma_\theta [ mr].\delta[10^{-3 } ] } { \epsilon_{6n}[10^{-12 } ] } \right ) ^{2/3 } } .\end{aligned}\ ] ] this formula uses the standard mcc assumption[1 ] that the ratio of transverse to longitudinal emittances can be chosen in the muon cooling channel to maximize the luminosity for a given @xmath2 .
the pinch enhancement factor , @xmath12 , is very close to unity ( see table 1 ) , and the numerical coefficient in equation 1 includes a geometric correction factor of 0.76 for the non - zero bunch length , @xmath13 ( the `` hourglass effect '' ) .
the physics motivation for each of the parameter sets in table 1 is discussed in [ 2 ] .
briefly , the number of @xmath14 events gives a benchmark estimate of the discovery potential for elementary particles at the full com energy of the collider , while the production of hypothesized 100 gev higgs particles indicates roughly how the colliders would perform in studying physics at this fixed energy scale .
further information on the important issue of neutrino radiation can be found in [ 3 ] .
the numbers given in table 1 come from an analytical calculation that is not intended to be accurate at much better than an order of magnitude level and that is deliberately conservative , i.e. it may well overestimate the radiation levels .
the radiation levels are predicted to rise approximately as the cube of the collider energy if other relevant parameters are held fixed ( up to some mitigating factors that come into play at the highest energies ) , rapidly becoming a serious design constraint for colliders at the tev scale and above .
the 1 tev parameter set of table 1 would give about the same luminosity as , for example , the design for the proposed nlc linear electron collider at the same energy , and the physics motivation and capabilities might be relatively similar[2,4 ] .
placement of the collider at 125 meters depth reduces the average neutrino radiation in the collider plane to less than one thousandth of the u.s .
federal off - site radiation limit ( 1 msv / year , which is of the same order of magnitude as the typical background radiation from natural causes ) .
nevertheless , attention would still need to be paid to minimizing the length , l , of any straight sections with low beam divergence , since these produce radiation hotspots with intensity proportional to l[3 ] .
the 4 tev parameter set was chosen as being at about the highest energy that is practical for a `` first generation '' muon collider on an existing laboratory site , due to neutrino radiation , and the muon current has been reduced to lower the radiation to the same level as the 1 tev parameter set , accepting the consequent loss in luminosity .
the 4 tev parameters may be compared to the mcc 4 tev design presented at snowmass96[5 ] , which did not take account of the neutrino radiation issue and hence attained a luminosity higher by more than an order of magnitude
. the lower bunch repetition rate of the current 4 tev parameter set makes some of the design parameters more relaxed than in the snowmass design , particularly in allowing a `` lite front end '' with relaxed rate specifications : the proton driver , pion production target and cooling channel . on the other hand , the desire to recover some of the lost luminosity motivates collider ring parameters that are slightly more aggressive , especially @xmath15 ( 3 mm reduced to 1.2 mm ) and @xmath8 ( 0.9 mrad increased to 1.6 mrad )
this entails a more difficult final focus design and also a more difficult task to shield the detector region from muon decay backgrounds . beyond com energies of a few tev
, it is probably necessary to build the colliders at isolated sites where the public would not be exposed to the neutrino radiation disk .
these will presumably be `` second generation '' machines , arriving after the technology of muon colliders has been established in one or more smaller and less expensive machines built at existing hep laboratories .
the gain from being able to relax the neutrino radiation constraint is evident in the 10 tev parameter set , with an exciting luminosity of @xmath16 at several times the discovery mass reach of the lhc hadron collider . from the progression of the parameter sets it is clear that the final focus design will become progressively more difficult with rising com energy .
consider , for example , the overall beam demagnification , @xmath17 , a dimensionless parameter that should be closely correlated with fractional tolerances in magnet uniformity , residual chromaticity etc .
for the 10 tev example , this has risen to approximately 31 000 in both the x and y coordinates , which as a very crude comparison that ignores considerable differences in the other final focus parameters happens to be approaching the y - coordinate value for both the 0.5 tev ( ia ) and 1.0 tev ( iia ) designs for the nlc linear collider ( i.e. 39 000 , with @xmath18 and @xmath19 ) [ 4 ] .
the spot size clearly indicative of vibration and alignment tolerances is also falling , but even at 100 tev it remains an order of magnitude above the spot size in the y coordinate for the nlc design parameters .
for perspective , then , the design of the final focus at 10 tev com energy may well still be less challenging than the design of the muon cooling channel , and the latter task is essentially independent of the collider energy ( up to assumed advances for later generation colliders ) .
the highest energy parameter set in table 1 , at 100 tev , clearly presents the most difficult design challenge , for several reasons : 1 ) cost reductions will be needed to make a machine of this size affordable , 2 ) siting will be more difficult than at 10 tev , since the neutrino radiation is now well above the u.s .
federal limit , 3 ) @xmath17 is almost an order of magnitude larger than at 10 tev , 4 ) the assumed @xmath20 is 25 times smaller than for the 10 tev parameters , albeit with much smaller bunches , so the assumed phase space density is nearly a factor of two larger , and finally 5 ) the beam power has risen to 170 mw , with synchrotron radiation rising rapidly to contribute a further 110 mw .
most of these extrapolations correspond to incremental advances in technology , particularly involving magnets : magnetic field strength ( for improved cooling and final focus , smaller accelerating rings and collider rings ) , stability and uniformity ( particularly for the final focus ) and cost reduction ( for the accelerating rings and collider rings ) .
hence , it is certainly not ruled out that such a parameter set could become achievable after a couple of decades of research and development dedicated to muon collider technology .
it has been shown that muon collider parameter sets at up to 10 tev com energy may well be realistic by today s standards of technology while muon colliders at the 100 tev energy scale require technological extrapolations that could perhaps be achievable within the relatively near - term future .
[ 1 ] the muon collider collaboration , `` status of muon collider research and development and future plans '' , to be submitted to phys
. rev . e. [ 2 ] b.j .
king , `` muon colliders : new prospects for precision physics and the high energy frontier '' , bnl cap-224-muon-98c , submitted to proc .
latin am . symp . on hep ,
april 8 - 11,1998 , san juan , puerto rico , ed .
available at http://pubweb.bnl.gov / people / bking/. [ 3 ] b.j .
king , `` a characterization of the neutrino - induced radiation hazard at tev - scale muon colliders '' , bnl cap-162-muon-97r , to be submitted for publication . [ 4 ] the nlc design group , `` zeroth - order design report for the next linear collider '' , lbnl - pub-5424 , slac report 474 , ucrl - id-124161 , may 1996 .
.self - consistent parameter sets for muon colliders .
the generation of these parameter sets is discussed in the text . except for the first parameter set , which has been studied in some detail by the muon collider collaboration , these parameters represent speculation by the author on how muon colliders might evolve with energy .
the beam parameters at the interaction point are defined to be equal in the horizontal ( x ) and vertical ( y ) transverse coordinates . [ cols=">,^,^,^,^,^ " , ] | some of the potential capabilities and design challenges of muon colliders are illustrated using self - consistent collider parameter sets at center of mass energies ranging from 0.1 tev to 100 tev . |
a _ matching _ in a graph is a subset of pairwise disjoint edges . a bipartite graph @xmath0 with at least one edge is said to be _ identifiable _ if for every vertex in @xmath4 , the subgraph of @xmath3 induced by its non - neighborhood has a matching of cardinality @xmath2 .
identifiable bipartite graphs were studied in several papers @xcite ; the property arises in the context of low - rank matrix factorization and has applications in data mining , signal processing , and computational biology . for further details on applications of notions and problems discussed in this paper
, we refer to @xcite . while the recognition problem for identifiable bipartite graphs is clearly polynomial using bipartite matching algorithms , several natural algorithmic problems concerning identifiable graphs turn out to be np - complete ( see @xcite ) . in @xcite , three problems related to finding specific identifiable subgraphs
were introduced . to state these problems ,
we need to recall the notion of an @xmath5-subgraph of a bipartite graph ( which appeared first in @xcite and , in a slightly modified form , which we will adopt , in @xcite ) . for a bipartite graph @xmath0 and vertex sets @xmath8 , @xmath9 ,
we denote by @xmath10 $ ] the subgraph of @xmath3 induced by @xmath11 .
let @xmath0 be a bipartite graph .
for a subset @xmath12 , the _ @xmath5-subgraph of @xmath3 induced by @xmath13 _ is the subgraph @xmath14 $ ] , where @xmath15 denotes the set of all vertices in @xmath16 with a neighbor in @xmath17 .
we say that a graph @xmath18 is an _
@xmath5-subgraph of @xmath3 _ if there exists a subset @xmath12 such that @xmath19 .
the following three problems are all related to finding identifiable @xmath5-subgraphs of a given graph : in @xcite , the optimization version of the min - identifiable subgraphproblem was shown to be @xmath6-hard . in the same paper it was shown that all three problems are polynomially solvable for trees , as well as for bipartite graphs @xmath20 such that the maximum degree of vertices in @xmath4 is at most @xmath21 . in @xcite , restricted versions of the identifiable subgraphproblem were studied , parameterizing the instances according to the maximum degree @xmath22 of vertices in @xmath16 . formally : it was shown in @xcite that the @xmath23-bounded identifiable subgraph problem for @xmath24 is as hard as identifiable subgraphproblem in general and that the @xmath21-bounded identifiable subgraph problem is solvable in linear time .
the complexity of the identifiable subgraphand max - identifiable subgraphproblems in general bipartite graphs was left open by previous works . in this paper , we establish the computational complexity of the identifiable subgraphand max - identifiable subgraphproblems , showing that both problems are solvable in polynomial time .
the key idea to our approach is the observation that if the input graph @xmath20 is not identifiable , then one can compute in polynomial time a maximal subset @xmath25 no vertex of which is contained in any identifiable @xmath5-subgraph of @xmath3 .
such a set @xmath26 is non - empty and can be safely deleted from the graph together with all its neighbors , thus reducing the problem to a smaller graph .
if the algorithm finds an identifiable @xmath5-subgraph of @xmath3 , then it in fact finds an identifiable @xmath5-subgraph of @xmath3 induced by a largest possible subset of @xmath4 , thereby also solving the max - identifiable subgraphproblem .
the proof also shows that such a subgraph is unique . in the second part of the paper , we complement the @xmath6-hardness result for the optimization version of the min - identifiable subgraphproblem from @xcite by studying the problem from the parameterized complexity point of view .
we introduce two natural parameterized variants of the min - identifiable subgraphproblem and prove that both are @xmath7}$]-hard , by giving parameterized reductions from the well - known @xmath7}$]-hard multicolored clique(@xmath23)problem .
the paper is structured as follows . in section [ sec : prelim ]
, we give the necessary definitions . in section [ sec : algorithm ] , we give a polynomial time algorithm that simultaneously solves the identifiable subgraphand the max - identifiable subgraphproblems . in section [ sec : parameterized ] , we study the @xmath27-hard min - identifiable subgraphproblem from the parameterized complexity point of view .
section [ sec : conclusion ] concludes the paper with some open questions .
all graphs considered in this paper are finite , simple , and undirected . for a graph @xmath3 ,
we denote by @xmath28 the vertex set of @xmath3 and by @xmath29 its edge set .
bipartite graph _ is a graph @xmath30 such that there exists a partition of @xmath31 into two sets @xmath4 and @xmath16 such that @xmath32 and @xmath33 . in this paper
, we will regard bipartite graphs as already _ bipartitioned _ , that is , given together with a fixed bipartition @xmath34 of their vertex set , and hence use the notation @xmath20 . for a graph @xmath35 and a subset of vertices @xmath36
, @xmath37 denotes the neighborhood of @xmath38 , i.e. , the set of all vertices in @xmath39 that have a neighbor in @xmath38 . for a vertex @xmath40 , we write @xmath41 for @xmath42 , and denote the _ degree _ of @xmath43 with @xmath44 . in @xmath37 , @xmath41 , @xmath45
, we shall omit the subscript @xmath3 if the graph is clear from the context .
a _ clique _ in a graph is a set of pairwise adjacent vertices .
a _ parameterized problem _ is a language @xmath46 ; the second component , @xmath23 , of instances @xmath47 is called the _ parameter_. a parameterized problem @xmath48 is _ fixed - parameter tractable _ ( fpt ) if there is a function @xmath49 , a constant @xmath50 , and an algorithm @xmath51 that decides @xmath52 in time @xmath53 for all @xmath47 .
let @xmath54 denote the class of all fixed - parameter tractable parameterized problems .
a _ parameterized reduction _ from @xmath46 to @xmath55 is a mapping @xmath56 such that there are functions @xmath57 and a constant @xmath50 with : @xmath52 if and only if @xmath58 , the parameter value @xmath59 of @xmath60 is at most @xmath61 , and @xmath62 can be computed in time @xmath63 .
it is well known that the existence of a parameterized reduction from @xmath48 to @xmath64 and @xmath65 imply that @xmath66 as well , and that parameterized reducibility is transitive .
accordingly , similarly to @xmath67 vs. @xmath68 , there are hardness classes of problems that are suspected not to be fpt . in particular , it is believed that @xmath69}}\xspace}$ ] and , under this assumption , a parameterized reduction from any @xmath7}$]-hard problem rules out fixed - parameter tractability .
( here @xmath7}$]-hardness is with respect to parameterized reductions . ) for graph - theoretic definitions not given in the paper we refer to @xcite , for further background in matching theory to @xcite , and for background in parameterized complexity to @xcite .
in this section we give a polynomial time algorithm for the identifiable subgraphproblem , the problem of determining whether a given graph @xmath0 has an identifiable @xmath5-subgraph . as a corollary of our approach
we will also obtain a polynomial time algorithm for the max - identifiable subgraphproblem .
the key ingredient for the algorithm is the following lemma .
[ lemma : ptime : key ] let @xmath0 be a non - identifiable bipartite graph with at least one edge and let @xmath1 such that there is no matching of @xmath70 into @xmath71 .
let @xmath26 be an inclusion - wise minimal subset of @xmath70 that has no matching into @xmath71 .
such a set @xmath26 is nonempty and always exists .
moreover , no identifiable @xmath5-subgraph of @xmath3 contains a vertex of @xmath26 .
let @xmath26 be a minimal subset of @xmath70 that has no matching into @xmath72 . by hall s theorem there must be a subset @xmath73 with @xmath74 .
any such set @xmath75 has no matching into @xmath71 .
because @xmath26 is a minimal set without a matching into @xmath71 it follows that @xmath76 .
furthermore , every proper subset of @xmath26 does have a matching into @xmath71 .
now , fix an arbitrary set @xmath12 such that the induced @xmath5-subgraph @xmath77 $ ] is identifiable .
we need to show that @xmath78 .
assume for contradiction that @xmath79 .
let @xmath80 and @xmath81 . because @xmath82 we have that @xmath83 is a proper , possibly empty , subset of @xmath26 .
hence , by the first paragraph , we have that @xmath84 . in the @xmath5-subgraph @xmath85 induced by @xmath13 , by definition ,
none of the neighbors of @xmath83 are present .
thus , the vertices in @xmath86 have at most those vertices as neighbors that are adjacent to @xmath86 but not to @xmath83 .
( further vertices in @xmath87 may imply that further neighbors of @xmath86 are not present , but this will not be important . ) thus , the number of neighbors that @xmath86 has in the vertices of @xmath71 that are present in @xmath85 is at most @xmath88 it follows immediately that @xmath86 has no matching into @xmath71 in @xmath85 . if @xmath89 then testing the identifiability condition for @xmath90 would require such a matching . if @xmath91 then using that @xmath85 must have a matching of @xmath13 into @xmath92 means that we would need a matching of @xmath86 into @xmath93 .
thus , either way we get a contraction .
this implies that @xmath78 , as claimed .
given a graph @xmath3 and vertex @xmath90 as in lemma [ lemma : ptime : key ] the set @xmath26 can be found in a straightforward way by folklore knowledge about bipartite matchings .
we sketch a very simple algorithm by self - reduction for completeness .
[ lemma : ptime : minimalset ] given a non - identifiable graph @xmath0 and vertex @xmath1 such that there is no matching of @xmath70 into @xmath71 , a minimal set @xmath26 as in lemma [ lemma : ptime : key ] can be found in polynomial time .
set @xmath94 and repeat the following routine : try each vertex @xmath95 and test whether there is a matching of @xmath96 into @xmath71 .
if there is then try the next vertex . if not then update @xmath97 and repeat .
output the current set @xmath26 if each @xmath96 has a matching of @xmath96 into @xmath71 . as an invariant
, the set @xmath26 never has a matching into @xmath71 .
in particular , we can never reach an empty set ( and we can only reach a singleton vertex if it is isolated ) . thus , the algorithm must terminate with a nonempty set @xmath26 such that each set @xmath96 has a matching into @xmath71 .
this also means that all smaller subsets of @xmath26 have matchings into @xmath71 .
thus , @xmath26 is a minimal set with no matching into @xmath71 .
we know now that if @xmath0 is not identifiable then we can efficiently find a subset @xmath25 such that no vertex of @xmath26 is contained in any identifiable @xmath5-subgraph of @xmath3 .
we now prove formally that we may safely delete @xmath26 and @xmath98 from @xmath3 while still retaining the same set of identifiable @xmath5-subgraphs .
[ lemma : lsubgraph : deletion ] let @xmath0 a bipartite graph and let @xmath25 .
every @xmath5-subgraph of @xmath3 that contains no vertex of @xmath26 is also an @xmath5-subgraph of the @xmath5-subgraph of @xmath3 induced by @xmath99 , and vice versa .
moreover , these @xmath5-subgraphs are induced by the same sets @xmath100 .
every @xmath5-subgraph of a graph is defined by the left part of its bipartition .
we show that taking the induced @xmath5-subgraph for any @xmath100 gives the same graph from both @xmath3 and @xmath101=g[l\setminus k , r\setminus n(k)]$ ] .
fix an arbitrary set @xmath100 .
clearly , since @xmath13 is a subset of the left part of the bipartition in both graphs , we get @xmath5-subgraphs of the form @xmath102 $ ] and @xmath103)[j , r_2]$ ] .
the latter is also an induced subgraph of @xmath3 so it simplifies to @xmath104 $ ] .
it suffices to prove that @xmath105 . by definition of @xmath5-subgraph
we have @xmath106 .
similarly , for the @xmath5-subgraph of @xmath13 in @xmath107=g[l\setminus k , r\setminus n_g(k)]$ ] we get @xmath108 we can safely replace @xmath109 by @xmath110 because @xmath18 is an induced subgraph of @xmath3 so the neighborhood is only affected by restriction to @xmath111 , the right part of the bipartition of @xmath18 .
thus , in @xmath112 we have the vertices of @xmath16 that do not have a neighbor in @xmath26 and that do not have a neighbor in @xmath113 . because @xmath26 and @xmath13 are disjoint subsets of @xmath4 this is the same as taking out the neighbors of @xmath17 from @xmath16 , i.e. , taking @xmath114 .
thus , both graphs are induced subgraphs of @xmath3 with left part @xmath13 and right part @xmath115 , so they are identical as claimed .
in particular , the lemma implies that if no identifiable @xmath5-subgraph contains a vertex of a nonempty set @xmath25 then @xmath3 and @xmath101 $ ] contain the same identifiable @xmath5-subgraphs .
thus , when seeking identifiable @xmath5-subgraphs it is safe to eliminate @xmath116 $ ] for sets @xmath26 obtained via lemma [ lemma : ptime : key ] .
now we can put together the claimed polynomial time algorithm .
[ theorem : ids : ptime ] the identifiable subgraphproblem can be solved in polynomial time . the algorithm works as follows .
given an input graph
@xmath0 it first tests if @xmath117 . if @xmath117 , then @xmath3 is not identifiable and has no identifiable @xmath5-subgraph ; the algorithm reports this fact and halts .
if @xmath118 , the algorithm proceeds iteratively .
identifiability can be efficiently tested by @xmath119 bipartite matching computations .
if @xmath3 is identifiable then graph @xmath3 is output as an identifiable @xmath5-subgraph .
if @xmath3 is not identifiable then the algorithm picks an arbitrary @xmath90 such that there is no matching of @xmath70 into @xmath71 . by lemma [ lemma : ptime : key ]
there is a nonempty set @xmath25 such that no identifiable @xmath5-subgraph of @xmath3 contains a vertex of @xmath26 ; such a set can be found efficiently by lemma [ lemma : ptime : minimalset ] .
thus , if @xmath3 has any identifiable @xmath5-subgraph then every such subgraph must avoid @xmath26 and , hence , it is also an @xmath5-subgraph of @xmath101 $ ] by lemma [ lemma : lsubgraph : deletion ] .
conversely , @xmath101 $ ] contains no further identifiable @xmath5-subgraphs .
the algorithm thus replaces @xmath3 by @xmath101 $ ] and starts over . in case
a graph is output , lemma [ lemma : lsubgraph : deletion ] implies that the output graph is also an @xmath5-subgraph of the initial graph @xmath3 .
in fact , it can be easily seen that the algorithm always returns a maximum identifiable @xmath5-subgraph and thus also solves the maximization variant of the problem , max - identifiable subgraph , in polynomial time .
( the proof also shows that this graph is unique . )
[ corollary : max - ids : ptime ] the max - identifiable subgraphproblem can be solved in polynomial time .
clearly , if the input graph is identifiable then returning it is optimal . if not then either @xmath117 ( in which case @xmath3 has no identifiable @xmath5-subgraph ) or there is a vertex @xmath90 such that @xmath70 can not be matched into @xmath71 and , by lemma [ lemma : ptime : key ] the algorithm finds a nonempty set @xmath26 that is avoided by all identifiable @xmath5-subgraphs . since @xmath3 and @xmath101 $ ] have the same @xmath5-subgraphs induced by @xmath100 ,
in particular , any maximum identifiable @xmath5-subgraph of @xmath3 is also an identifiable @xmath5-subgraph of @xmath101 $ ] .
thus , continuing the iterative approach on @xmath101 $ ] will find a maximum solution , if one exists .
in this section we study the parameterized complexity of the min - identifiable subgraphproblem , which was proved @xmath27-hard in a previous work @xcite .
we consider the following parameterized variants min - identifiable subgraph(@xmath23)and min - identifiable subgraph(@xmath120 ) .
the two problems differ only in the choice of parameter ; the min - identifiable subgraph(@xmath120)problem can be reformulated as the problem of finding a set @xmath121 of size at least @xmath23 such that @xmath122 $ ] is identifiable .
we show that both parameterizations are @xmath7}$]-hard , i.e. , they are not fixed - parameter tractable unless @xmath123}}\xspace}$ ] , which is deemed unlikely . for both problems
we give parameterized reductions from the well - known @xmath7}$]-hard multicolored clique(@xmath23)problem , defined as follows
. min - identifiable subgraph(@xmath23)is @xmath7}$]-hard .
we give a parameterized reduction from multicolored clique(@xmath23)to min - identifiable subgraph(@xmath23 ) .
let @xmath124 be an instance of multicolored clique(@xmath23 )
. without loss of generality assume that @xmath125 or else solve the instance in polynomial time ( finding a clique of size @xmath126 ) .
let @xmath127 for @xmath128 .
we will construct a bipartite graph @xmath129 such that @xmath130 is yes for multicolored clique(@xmath23)if and only if @xmath131 is yes for min - identifiable subgraph(@xmath23 ) .
_ construction .
_ we create an instance of min - identifiable subgraph(@xmath23)with bipartite graph @xmath129 and parameter @xmath132 .
the set @xmath4 consists of the vertices in @xmath31 along with @xmath23 special vertices @xmath133 .
we now describe the set @xmath16 along with the adjacencies between @xmath4 and @xmath16 : * for each choice of @xmath134 we create a set @xmath135 of vertices , which is then added to @xmath16 . *
* for each edge @xmath136 in @xmath3 with @xmath137 and @xmath138 we add a vertex to @xmath135 and make it adjacent to @xmath139 and @xmath90 in @xmath18 .
( we could also achieve this by starting with @xmath3 , dropping the ( irrelevant ) edges between vertices of the same set @xmath140 , and then subdividing every edge . ) * * make the special vertex @xmath141 adjacent to all vertices of @xmath142 with @xmath143 and @xmath144 .
* for each @xmath145 create a set @xmath146 of @xmath147 vertices and add it to @xmath16 . * * make each vertex @xmath148 adjacent to @xmath23 private vertices in @xmath146 .
no other vertices of @xmath31 will be adjacent to these vertices . *
* make each special vertex @xmath149 adjacent to all vertices of @xmath146 with @xmath150 .
this completes the construction of @xmath129
. it can be helpful to keep in mind that vertices in @xmath140 are only adjacent to ( some ) vertices in @xmath146 or in @xmath142 with @xmath151 , whereas each special vertex @xmath141 is adjacent to all vertices in @xmath152 with @xmath150 and all vertices in @xmath142 with @xmath153 .
an example construction is shown in fig . [
fig : reduction ] . from an input @xmath130 to multicolored clique(@xmath23 ) . in the example @xmath154 ,
vertices in @xmath4 and @xmath16 are colored black and white , respectively , and a thick edge between vertex @xmath141 and a set @xmath152 means that @xmath141 is adjacent to all vertices in @xmath152.,width=604 ] clearly , the construction can be performed in polynomial time .
it remains to prove correctness , that is , that @xmath3 has a @xmath23-clique containing exactly one vertex of each set @xmath140 if and only if @xmath131 is a yes instance of min - identifiable subgraph(@xmath23)for @xmath132 .
_ correctness .
_ assume first that @xmath3 contains a @xmath23-clique @xmath155 with exactly one vertex from each set @xmath140 and let @xmath156 . we claim that @xmath157 induces an identifiable @xmath5-subgraph
let @xmath158 denote the vertices in @xmath16 of the induced @xmath5-subgraph , i.e. , the vertices that have no neighbor among @xmath159 .
let us first check that there is a matching of @xmath160 into @xmath161 : the vertex @xmath141 is not adjacent to @xmath146 nor to sets @xmath142 with @xmath153 .
there are @xmath23 vertices in @xmath146 that are adjacent to @xmath162 as well as the vertex @xmath163 .
these are contained in @xmath164 since all their neighbors are in @xmath165 .
we can match the mentioned vertices of @xmath165 to them . because @xmath155 is a clique
there are edges from @xmath163 to each other vertex of the clique ; these give rise to vertices in @xmath166 corresponding to these edges that have no other neighbors in @xmath167 ( in @xmath18 ) .
because @xmath141 is not adjacent to such sets @xmath142 all these vertices are present and each @xmath168 can be matched to the vertex representing its edge to @xmath163 .
let us now check that there is a matching of @xmath169 into @xmath170 : for each @xmath171 with @xmath172 all its neighbors in @xmath152 are present since they have no other neighbor in @xmath31 and all @xmath133 are in the @xmath5-subgraph .
thus , each @xmath171 can be matched to such a neighbor . because @xmath125 there are at least two sets @xmath152 and @xmath173 with @xmath174 and a total of @xmath175 vertices therein are not yet matched to .
thus , all vertices @xmath133 can be matched to these vertices .
( e.g. , all but @xmath149 to vertices of @xmath152 and @xmath149 to a vertex of @xmath173 . )
thus , the @xmath5-subgraph induced by @xmath165 is indeed identifiable . since @xmath176 this implies that @xmath177 is a yes instance of min - identifiable subgraph(@xmath23 )
now assume that @xmath177 is yes for min - identifiable subgraph(@xmath23 ) .
let @xmath178 be a set of size at most @xmath179 such that the @xmath5-subgraph of @xmath18 induced by @xmath165 , namely @xmath180 $ ] with @xmath181 , is identifiable .
our goal is to show that it includes all special vertices @xmath133 along with one vertex per set @xmath140 and that the latter vertices form a @xmath23-clique in @xmath3 .
we first observe that @xmath182 and @xmath183 is required : excluding either type of vertex implies @xmath184 since each vertex in @xmath16 is adjacent to at least one vertex of @xmath31 and at least one vertex @xmath141 .
assume for contradiction that at least two special vertices , say @xmath141 and @xmath149 with @xmath150 , are not in @xmath165 .
thus , picking a third vertex @xmath185 we need a matching of @xmath186 into the vertices of @xmath16 that are not neighbors of ( at least ) @xmath141 , @xmath149 , and @xmath187 , but no such vertices exist : there is no @xmath188 with @xmath189 and @xmath190 and there is no @xmath142 with @xmath151 , @xmath191 , @xmath192 . since we have at least one vertex @xmath193 we can observe that this vertex can not be matched ; a contradiction .
now assume for contradiction that exactly one special vertex , say @xmath194 , is not contained in @xmath165 .
this requires a matching of @xmath195 into @xmath196 . since both @xmath197 , no vertex of a set
@xmath142 may exist in @xmath196 except possibly for vertices of @xmath198 .
thus , no vertex of @xmath199 may be in @xmath165 since they would have no neighbors to match to .
this in turn implies that no other set @xmath142 except for @xmath198 has any vertices in @xmath164 .
we complete the contradiction by considering the requirement of a matching of @xmath200 into @xmath201 : now , the vertices of @xmath198 are not available since they are adjacent to @xmath202 .
thus , there are no @xmath142 vertices to match to . similarly ,
absence of @xmath194 and @xmath202 eliminates all vertices of sets @xmath188 .
since @xmath203 must contain at least one vertex of @xmath31 , we find that such a vertex can not be matched into @xmath201 ; a contradiction .
we now have the remaining case that @xmath204 .
we also know already that at least one vertex of @xmath31 must be contained in @xmath165 , say @xmath205 and pick @xmath206 .
assume for contradiction that some set @xmath140 with @xmath207 has an empty intersection with @xmath165 .
it follows that in @xmath164 there are no vertices of sets @xmath142 with @xmath151 since each such vertex is adjacent to some vertex in @xmath140 .
we now consider the requirement of a matching of @xmath160 into @xmath161 to complete the contradiction : this additionally ensures that there are no vertices of @xmath208 left in @xmath161 as well as no vertices of @xmath142 with @xmath209 , implying that there are no neighbors for @xmath210 to match to ; a contradiction .
thus , we have @xmath204 and @xmath165 has a nonempty intersection with each set @xmath211 . because @xmath165 has size at most @xmath179 this directly implies that its size is exactly @xmath179 and that it contains exactly one vertex of each set @xmath140 , say @xmath212 .
it remains to show that @xmath213 is a clique in @xmath3 .
assume for contradiction that this is not the case , say that @xmath163 and @xmath171 for @xmath150 are not adjacent in @xmath3 .
consider the requirement of a matching of @xmath214 into @xmath161 : absence of @xmath141 ensures that no vertex of @xmath152 is present .
moreover , no vertices of @xmath142 for @xmath144 are in @xmath161 .
in particular , for vertex @xmath171 this only leaves vertices in @xmath135 or @xmath215 ( depending on whether @xmath216 or @xmath217 ) .
because @xmath163 and @xmath171 are not adjacent , however , and no other vertex of @xmath140 is in @xmath165 , no such vertices exist in @xmath218 .
thus , @xmath171 can not be matched ; a contradiction .
it follows that the vertices @xmath219 must indeed form a clique in @xmath3 .
this completes the proof .
the above proof also shows that the problem of testing whether a given bipartite graph has an identifiable @xmath5-subgraph induced by a set @xmath12 with @xmath220 is @xmath7}$]-hard ( with respect to parameter @xmath23 ) .
[ theorem : minidsnk ] min - identifiable subgraph(@xmath120)is @xmath7}$]-hard .
we give a parameterized reduction from multicolored clique(@xmath23)to min - identifiable subgraph(@xmath120 ) .
let @xmath124 be an instance of multicolored clique(@xmath23 ) .
w.l.o.g . , assume @xmath125 or else solve the instance in polynomial time .
let @xmath221 and let @xmath127 for @xmath128 .
assume w.l.o.g .
that each set @xmath140 contains at least two vertices ( else we can restrict the graph to the subgraph induced by the neighborhood of @xmath222 and drop color @xmath223 to get an equivalent instance ) .
we will construct a bipartite graph @xmath129 such that @xmath130 is yes for multicolored clique(@xmath23)if and only if @xmath131 is yes for min - identifiable subgraph(@xmath120)for @xmath224 .
that is , @xmath130 should be yes if and only if the graph @xmath18 contains an @xmath5-identifiable subgraph that is induced by a set @xmath165 of size at most @xmath120 .
note that @xmath131 has a parameter value of @xmath225 .
recall that we can equivalently ask for the existence of a set @xmath121 of size at least @xmath23 such that @xmath226 $ ] is identifiable since then @xmath227 is of size at most @xmath120 and can play the role of the requested set @xmath165 ( and conversely @xmath159 is a feasible choice for @xmath228 ) . define @xmath229 ; this value will be used in the construction .
_ construction .
_ the graph @xmath129 is defined as follows : * the vertex set @xmath4 consists of @xmath31 as well as a set @xmath230 of special vertices @xmath133 . *
the set @xmath16 contains for each vertex @xmath231 a set of @xmath232 vertices @xmath233 whose only neighbor in @xmath31 will be @xmath90 ( so they are in a limited sense a private neighbors of @xmath90 ) .
let @xmath146 denote the set of vertices @xmath234 with @xmath148 for each @xmath128 .
( the exact number of these vertices per vertex @xmath90 will be immaterial so long as they are at least @xmath232 . )
* the set @xmath16 furthermore contains vertices derived from the edges of @xmath3 .
let @xmath235 be any edge of @xmath3 with @xmath222 and @xmath236 .
create @xmath237 vertices @xmath238 and add them to @xmath16 .
make each of them adjacent to all vertices of @xmath239 and all vertices of @xmath240 .
do this for all edges for any @xmath134 and let @xmath135 contain the vertices @xmath238 for edges @xmath241 between @xmath140 and @xmath242 in @xmath3 .
the set @xmath16 is thus the union of sets @xmath188 for @xmath243 and sets @xmath142 for @xmath143 .
( again , the exact value of @xmath237 is not important so long as @xmath244 . )
* make each vertex @xmath245 adjacent to all vertices of each set @xmath188 with @xmath243 .
furthermore , make each @xmath141 adjacent to all vertices of each set @xmath142 with @xmath153 .
define @xmath224 and return the instance @xmath131 .
an example construction is shown in fig . [
fig : reduction-2 ] . from an input @xmath130 to multicolored clique(@xmath23 ) . in the example @xmath154 , vertices in @xmath4 and @xmath16 are colored black and white , respectively , and an edge between vertex @xmath246 and a set @xmath247 of vertices in @xmath16 means that @xmath139 is adjacent to all vertices in @xmath247.,width=642 ] clearly this construction can be performed in polynomial time and we already pointed out that the parameter value of @xmath131 is equal to @xmath23 .
( parameter value bounded by any function of @xmath23 would be enough for a parameterized reduction . )
it remains to prove correctness , that is , that @xmath3 has a @xmath23-clique containing exactly one vertex of each set @xmath140 if and only if @xmath248 is a yes instance of min - identifiable subgraph(@xmath120 ) .
_ correctness .
_ assume first that @xmath130 is yes for multicolored clique(@xmath23 ) .
thus , @xmath3 contains a clique @xmath155 containing exactly one vertex of each set @xmath140 .
we claim that @xmath249 induces an identifiable @xmath5-subgraph in @xmath18 .
the @xmath5-subgraph induced by @xmath165 is exactly @xmath250 $ ] where @xmath251 .
we need to show that @xmath18 is identifiable .
let us first see that @xmath252 has a matching of @xmath253 into @xmath254 for any @xmath255 .
fix any such vertex @xmath256 : recall that in the construction we made for each vertex @xmath231 vertices @xmath233 such that @xmath90 is their only neighbor in @xmath31 . for each vertex
@xmath257 all these vertices @xmath234 exist in @xmath258 since @xmath259 so we can match each @xmath90 to @xmath260 .
moreover , for any @xmath261 we can match all @xmath23 vertices of @xmath230 to @xmath262 .
( here we tacitly assume that @xmath3 has more than @xmath232 vertices , which is w.l.o.g . )
this completes the required matching .
let us now exhibit a matching in @xmath252 of @xmath214 into @xmath263 for any @xmath264 : note that @xmath265 in particular does not contain vertices of sets @xmath188 for @xmath243 nor vertices of @xmath142 for @xmath144 since all those are adjacent to @xmath141 in @xmath18 ( so they are also adjacent to @xmath141 in @xmath252 ) .
it remains to use vertices of @xmath142 with @xmath151 , recalling that many of them are not present already in @xmath266 .
fix @xmath172 .
let @xmath267 and @xmath268 . since @xmath155 is a clique , vertices @xmath163 and @xmath171 are adjacent in @xmath3 .
for the corresponding edge @xmath235 we created vertices @xmath238 in @xmath269 ( or @xmath215 if @xmath270 ) .
each @xmath271 is adjacent to all of @xmath239 and @xmath240 but not to @xmath163 or @xmath171 ; they are not adjacent to any vertex of @xmath272 .
thus , all of these vertices are present in @xmath164 and hence also in @xmath263 .
we can therefore match all vertices of @xmath239 , @xmath240 , and @xmath273 to them since these are in total less than @xmath274 vertices . by repeating the argument for all @xmath275 we can also match vertices in @xmath276 , obtaining a matching for all of @xmath277 .
( here we need that @xmath24 so that we can match @xmath149 . )
it follows that the @xmath5-subgraph induced by @xmath165 is indeed identifiable . since @xmath278 it follows that @xmath131 is yes for min - identifiable subgraph(@xmath120 ) .
assume now that @xmath131 is yes for min - identifiable subgraph(@xmath120)and let @xmath165 be a subset of @xmath4 of size at most @xmath224 such that the @xmath5-subgraph induced by @xmath165 , namely @xmath279=g'[l',r']$ ] where @xmath181 , is identifiable .
let @xmath280 .
this is a set of size at least @xmath23 .
we will show that @xmath155 is a subset of @xmath23 vertices of @xmath31 that form a clique in @xmath3 with @xmath281 for @xmath145 .
note that @xmath266 .
we begin with some observations : if @xmath282 then @xmath184 because each vertex of @xmath16 is adjacent to at least one vertex of @xmath31 . in this case
, @xmath252 could not be identifiable since that requires having at least one edge .
thus , @xmath283 and we pick an arbitrary vertex @xmath284 to be used later .
similarly , if @xmath285 then again @xmath184 and @xmath180 $ ] can not be identifiable .
we pick @xmath286 to be used later .
( note that @xmath287 for some @xmath145 . )
we will now prove several restrictions on @xmath155 by contradiction - based arguments .
the first two aim at proving that @xmath288 .
assume for contradiction that @xmath289 .
thus , @xmath252 being identifiable implies that there must be a matching of @xmath290 to @xmath291 .
say @xmath292 with @xmath150 , then @xmath291 contains no vertices of @xmath16 that are adjacent to any of @xmath293 , @xmath141 , or @xmath149 .
this is a contradiction since every vertex of @xmath16 is adjacent to at least one of them : vertices in any @xmath188 are adjacent to each vertex of @xmath230 and vertices in any @xmath142 are only not adjacent to two vertices of @xmath230 , namely @xmath294 and @xmath295 .
thus , @xmath296 .
assume for contradiction that @xmath297 and assume w.l.o.g .
that @xmath298 .
because @xmath252 is identifiable there must be a matching of @xmath195 into @xmath196 in @xmath252 .
note that in @xmath196 there is no vertex of any set @xmath188 .
similarly , vertices of @xmath142 are not in @xmath196 unless @xmath299 and @xmath300 since otherwise they are adjacent to at least one of @xmath194 or @xmath301 .
this in turn implies that @xmath165 contains no vertices from @xmath302 for @xmath303 since they have no neighbors in @xmath196 .
consequently , even @xmath304 contains no vertex of any set @xmath142 with @xmath305 because the vertices of other sets @xmath142 are all adjacent to some vertex of @xmath306 .
now , consider the requirement of a matching of @xmath203 into @xmath201 in @xmath252 .
we now get that there are no vertices of sets @xmath188 nor of sets @xmath142 .
the latter holds because in @xmath164 only vertices of @xmath198 can exist but all those are adjacent to @xmath202 and hence not in @xmath201 .
thus , @xmath201 is empty and we can not match the vertex @xmath307 anywhere ; a contradiction .
we now know that @xmath308 .
assume that @xmath155 contains at least two vertices of the same set @xmath140 , i.e. , that @xmath309 . let @xmath310 with @xmath311 .
crucially , for any @xmath143 with @xmath151 , all vertices in @xmath142 are adjacent to at least one of @xmath163 and @xmath312 , by construction : a vertex @xmath271 for @xmath313 with @xmath314 and @xmath315 is adjacent to all vertices of @xmath316 .
because only one of @xmath163 and @xmath312 can be equal to @xmath317 it follows that @xmath271 is adjacent to at least one of the two .
thus , no vertex of @xmath318 is present in @xmath319 for @xmath320 and @xmath321 with @xmath151 .
now , because @xmath252 must be identifiable there must be a matching of @xmath160 into @xmath263 .
there are no vertices of any set @xmath188 in @xmath265 because all of them are adjacent to @xmath141 .
for the same reason there are no vertices of sets @xmath142 when @xmath144 . from above
we know that for @xmath142 with @xmath151 no such vertices are present in @xmath322 .
thus , vertex @xmath284 can not be matched ; a contradiction . at this point we know that @xmath155 contains no vertices of @xmath230 and at most one vertex of each set @xmath140 .
because the size of @xmath155 is at least @xmath23 this implies that @xmath155 contains exactly one vertex of each set @xmath140 and no further vertices ( it is of size exactly @xmath23 ) .
let @xmath156 for @xmath145 .
it remains to show that the vertices @xmath219 form a clique in @xmath3 .
assume for contradiction that @xmath222 and @xmath236 are not adjacent in @xmath3 for some @xmath150 ; w.l.o.g . @xmath216 .
again the construction of the edge - related vertices @xmath271 is important here : consider any edge @xmath323 with @xmath324 and @xmath325 .
s(here we also tacitly assume that there is such an edge , which is w.l.o.g .
as otherwise @xmath130 is a no instance to multicolored clique(@xmath23 ) . ) because there is no edge between @xmath163 and @xmath171 , we must have @xmath326 or @xmath327 ( or both ) .
this directly implies that each vertex @xmath328 is adjacent to @xmath163 or @xmath171 ( or both ) because it is adjacent to all of @xmath329 and all of @xmath330 .
it follows that there are no vertices of @xmath135 in @xmath331 . because @xmath252 is identifiable
there must be a matching of @xmath160 into @xmath161 . in the latter set
there are no vertices of any set @xmath188 because they are subsets of @xmath332 and similarly no vertices of @xmath142 if @xmath153 .
consider any vertex @xmath333 ( using that @xmath334 and @xmath335 ) : its neighbors in @xmath18 are in @xmath152 and in sets @xmath142 with @xmath191 .
this leaves only the set @xmath135 since we need @xmath151 and @xmath191 or else none of the vertices are in @xmath164 , but we already know that no vertices of @xmath135 are in @xmath336 ; a contradiction .
it follows that the vertices of @xmath155 form a clique in @xmath3 , with exactly one vertex from each set @xmath140 ; this proves that @xmath130 is yes for multicolored clique(@xmath23)and completes the proof .
in this paper , we showed that the identifiable subgraphand the max - identifiable subgraphproblems are polynomially solvable and that two natural parameterized variants of the min - identifiable subgraphproblem are @xmath7}$]-hard . regarding approximation issues , the min - identifiable subgraphproblem was shown to be @xmath6-hard @xcite , however , its exact ( in)approximability status remains an open question . in @xcite ,
two other @xmath27-hard problems related to identifiability were studied : finding the minimum number of edges that one must delete from a given identifiable graph to destroy identifiability and finding the smallest size of a set @xmath158 such that the graph @xmath337 $ ] is identifiable .
the hardness proof for the former problem shows that the problem is also @xmath7}$]-hard with respect to its natural parameterization .
more precisely , one can combine the @xmath27-hardness proof from @xcite and the proof of the @xmath27-hardness of the problem from which that reduction was made ( finding the minimum number of edges that one must delete from a given bipartite graph in order to decrease its matching number ) ( * ? ? ?
* theorems 3.2 and 3.3 and their proofs ) to obtain a parameter - preserving reduction from the parameterized clique problem , with respect to its natural parameterization , which is @xmath7}$]-hard .
we leave for future research the determination of the parameterized complexity status of the latter problem , as well as the ( in)approximability status of both problems .
this work was supported in part by the slovenian research agency ( i@xmath338-@xmath339 , research program p@xmath340-@xmath341 and research projects n@xmath340-@xmath342 , j@xmath340-@xmath343 , j@xmath340-@xmath344 , j@xmath340-@xmath345 , and j@xmath340-@xmath346 ) .
part of this research was carried out during the visit of m.m . to s.k . at university of bonn
; their hospitality and support is gratefully acknowledged .
e. fritzilas , y. a. rios - solis , and s. rahmann .
structural identifiability in low - rank matrix factorization . in _ computing and combinatorics _ ,
volume 5092 of _ lecture notes in comput .
_ , pages 140148 .
springer , berlin , 2008 .
l. lovsz and m. d. plummer .
, volume 121 of _ north - holland mathematics studies_. north - holland publishing co. , amsterdam ; akadmiai kiad ( publishing house of the hungarian academy of sciences ) , budapest , 1986 .
annals of discrete mathematics , 29 . | a bipartite graph @xmath0 with at least one edge is said to be _ identifiable _ if for every vertex @xmath1 , the subgraph induced by its non - neighbors has a matching of cardinality @xmath2 .
an of @xmath3 is an induced subgraph of @xmath3 obtained by deleting from it some vertices in @xmath4 together with all their neighbors .
the identifiable subgraph problem is the problem of determining whether a given bipartite graph contains an identifiable @xmath5-subgraph .
we show that the identifiable subgraph problem is polynomially solvable , along with the version of the problem in which the task is to delete as few vertices from @xmath4 as possible together with all their neighbors so that the resulting @xmath5-subgraph is identifiable .
we also complement a known @xmath6-hardness result for the complementary problem in which the task is to minimize the number of remaining vertices in @xmath4 , by showing that two parameterized variants of the problem are @xmath7}$]-hard . |
hadronic interactions of two photons lead to the production of hadrons whose properties in quantum chromodynamics ( qcd ) depend on the underlying partonic processes , as well as on the way in which the partons are transformed into observable hadrons .
the study of inclusive hadron production is therefore an appropriate tool to investigate the validity of qcd in hadronic photon - photon interactions .
furthermore a drastic discrepancy between data and nlo qcd has recently been observed in this process @xcite which would indicate the breakdown of this otherwise so successful theory and hence requires further study .
our measurements complement similar studies of jet production @xcite .
the same partonic processes are at work in qcd in both cases , but hadronisation is usually treated differently .
to predict the production of single hadrons , universal fragmentation functions obtained from fits to independent data are utilised to describe the transition from partons to hadrons .
jet observables on the other hand are designed to keep the influence of hadronisation processes small . the remaining distortion with respect to partonic jet observables
is estimated using hadronisation models .
the reliable calculation of an observable in perturbative qcd requires a sufficiently high energy scale which in our case is provided by the transverse momenta of the charged particles .
first results have been published by opal at @xmath1 centre - of - mass energies @xmath92 and @xmath93 gev @xcite .
l3 has published results in the range @xmath94 gev @xcite .
this paper extends the opal measurements of the transverse momentum spectra and pseudorapidity axis points towards the centre of the lep ring , the @xmath95 axis points upwards and the @xmath96 axis points in the direction of the electron beam .
the polar angle @xmath97 , the azimuthal angle @xmath98 and the radius @xmath99 denote the usual spherical coordinates .
the pseudorapidity @xmath100 is defined as @xmath101 . ]
distributions of charged hadrons in @xcite using data taken at @xmath1 centre - of - mass energies from @xmath102 to @xmath3 gev , representing a roughly thirty - fold increase in integrated luminosity .
the interactions of the photons can be modelled by assuming that each photon can either interact directly or appear resolved through its fluctuations into hadronic states . in leading order qcd
this model leads to three different event classes for @xmath103 interactions : direct , single - resolved and double - resolved , where resolved means that the partons ( quarks or gluons ) inside the hadronic photon take part in the hard interaction .
the probability to find partons in the photon is parametrised by parton density functions . at lep the photons
are radiated by the beam electrons and carry mostly small negative four - momenta squared , @xmath104 . in this paper events
are considered only if the electrons are scattered at small angles and are not detected .
both photons are therefore quasi - real ( @xmath105 gev@xmath106 ) .
differential hadron production cross - sections are measured as a function of the transverse momentum and the pseudorapidity of charged hadrons .
the inclusive cross - sections are compared to monte carlo ( mc ) event generators in leading order @xmath107 and theoretical calculations in next - to - leading order ( nlo ) @xmath107 for this process .
a detailed description of the opal detector can be found elsewhere @xcite .
the central tracking was performed inside a solenoidal magnet which provided a uniform axial magnetic field of 0.435 t along the beam axis . starting with the innermost components , the tracking system consisted of a high precision silicon microvertex detector , a precision vertex drift chamber , a large volume jet chamber with 159 layers of axial anode wires and a set of @xmath96 chambers measuring the track coordinates along the beam direction .
the transverse momenta , @xmath108 , of tracks are measured with a precision parametrised by @xmath109 ( @xmath108 in gev ) in the central region @xmath110 . in this paper transverse
is always defined with respect to the beam axis .
the magnet was surrounded in the barrel region ( @xmath111 ) by a lead glass electromagnetic calorimeter ( ecal ) and a hadronic sampling calorimeter ( hcal ) , which in turn were surrounded by muon chambers .
similar layers of detectors were installed in the endcaps ( @xmath112 ) .
the small angle region from 47 to 140 mrad around the beam pipe on both sides of the interaction point was covered by the forward calorimeters ( fd ) and the region from 33 to 59 mrad by the silicon tungsten luminometers ( sw ) .
the latter were used to determine the luminosity by counting small - angle bhabha scattering events .
the properties of the two interacting photons ( @xmath113 ) are described by their negative four - momentum transfers @xmath114 and their invariant mass .
each @xmath115 is related to the electron scattering angle @xmath116 relative to the beam direction by @xmath117 where @xmath118 and @xmath119 are the four - momenta of the beam electron and the scattered electron , respectively , and @xmath120 and @xmath121 are their energies
. events with detected scattered electrons ( single - tagged or double - tagged events ) are excluded from the analysis .
driven by the angular acceptance of the fd and sw calorimeters a value of @xmath122 gev@xmath106 is used in this analysis to separate the quasi - real photons of untagged events from the virtual photons of tagged events .
the median @xmath104 resulting from this definition can not be determined from data since the scattered electrons are not tagged . for the kinematic range of this analysis
the mc simulations predict the median @xmath104 to be of the order of @xmath123 gev@xmath106 .
the hadronic invariant mass of the photon - photon system , @xmath124 , can be obtained from the energies and momenta @xmath125 of final state particles : @xmath126 the mc generators pythia 5.722 @xcite and phojet 1.10 @xcite have been used to simulate photon - photon interactions .
pythia uses the sas-1d parametrisation @xcite for the photon parton densities and phojet uses the grv parametrisation @xcite .
all relevant background processes were studied using mc generators .
multihadronic events ( @xmath127 ) were simulated with pythia 6.125 .
koralz 4.02 @xcite was used to generate the process @xmath128 and vermaseren @xcite to generate @xmath129 .
deep - inelastic e@xmath130 events were simulated with herwig 5.9 @xcite .
all signal and background mc samples were generated with full simulation of the opal detector @xcite .
they are analysed using the same reconstruction algorithms as for the data .
the production of charged hadrons was studied using data taken at @xmath1 centre - of - mass energies , @xmath131 , from 183 to 209 gev , amounting to a total integrated luminosity of 612.8 pb@xmath132 .
the luminosity weighted average centre - of - mass energy is 195.8 gev .
two - photon events are selected with the following set of cuts : * at least six tracks must have been found in the tracking chambers . a track is required to have a minimum transverse momentum of 120 mev with respect to the @xmath96 axis and at least 40 hits in the central jet chamber .
the number of measured hits in the jet chamber must be more than half of the number of possible hits given the track direction .
the radial distance of nearest approach of the track to the primary vertex has to be less than 0.15 cm . *
the visible invariant hadronic mass calculated from the position and the energy of the clusters measured in the ecal has to be greater than 3 gev . *
the sum of all energy deposits in the ecal and the hcal has to be less than 50 gev to remove background from hadronic z decays in events with a radiative return to the z peak . * the missing transverse momentum of the event measured in the ecal and the fd has to be less than 8 gev . * to reject events with scattered electrons in the fd or sw , the total energy sum measured in the fd and sw has to be less than 60 gev . *
the background due to beam - gas or beam - wall interactions is reduced by requiring the radial distance of the primary vertex from the beam axis to be less than 2 cm and the distance from the nominal vertex position along the @xmath96 direction to be less than 3 cm .
after all cuts the data sample contains 1144035 events .
the main remaining background processes are multihadronic z decays , @xmath133 and deep - inelastic e@xmath130 scattering .
other background processes are found to be negligible .
the multihadronic background is mainly reduced by the cut on the sum of the energy measured by the ecal and the hcal and by the cut on the missing transverse momentum .
the @xmath133 processes are reduced by the cuts on the number of tracks and deep - inelastic e@xmath130 events are rejected by the cut on the energy in sw and fd .
the trigger efficiency for events in this region of phase space is close to 100% @xcite and no correction is applied . from the mc simulations
it is estimated that after all cuts the total remaining background is below 2% overall , but increasing to up to 50% at very high transverse momenta of the charged particles .
the background is subtracted bin - by - bin from the distributions measured before corrections for detector resolution and acceptance are applied .
the signal mc generators phojet and pythia are found to underestimate the cross - section . in particular they show a shape different from that in the data of the @xmath108-distribution of charged hadrons .
to study a potential bias resulting from this deficiency the mc events have been reweighted by a suitable @xmath108-dependent function to resemble the data .
both the original and the reweighted mc distributions are used in the analysis and any differences seen are included in the systematic uncertainty .
only particles with a lifetime @xmath134 ns are used to define the primary charged hadronic multiplicity .
the primary charged hadrons originate either directly from the primary interaction or from the decay of particles with a lifetime @xmath135 ns . the decay products of @xmath136 and @xmath137 particles are hence considered as primary hadrons in this analysis . the measured transverse momentum and pseudorapidity distributions of charged hadrons have to be corrected for the losses due to event and track selection cuts , the acceptance and the resolution of the detector .
this is done using the mc events which were processed by the full detector simulation and reconstruction chain .
the data are corrected by multiplying the experimental distribution , e.g. of the transverse momentum @xmath138 , with correction factors which are calculated as the bin - by - bin ratio of the generated and the reconstructed mc distributions : @xmath139 the ratio is calculated using both pythia and phojet , with the mean value used to correct the data .
the pseudorapidity distributions are corrected in the same way .
this method only yields reliable results if migration effects between bins due to the finite resolution of the measurement are small .
the bins of the @xmath108 and @xmath140 distributions have therefore been chosen to be significantly larger than the resolution expected from the mc simulation . to avoid regions where the detector has little or no acceptance , all measurements of charged hadron distributions
were restricted to the range @xmath141 .
the average transverse momentum @xmath142 and the average pseudorapidity @xmath143 in each bin is calculated directly from the data since detector corrections are small compared to the statistical uncertainties . in order to be able to use eq .
[ eq1 ] for the detector correction in each bin of @xmath124 , the bins in @xmath124 must be larger than the experimental resolution and the average reconstructed hadronic invariant mass , @xmath144 , should be approximately equal to the average generated hadronic invariant mass , @xmath145 . in this case
bin - to - bin migrations are minimised .
the visible invariant mass , @xmath146 , is determined from all tracks and calorimeter clusters , including the forward detectors and the silicon tungsten calorimeters , after applying a matching algorithm to avoid double counting of particle momenta @xcite .
the ratio of @xmath147 and @xmath146 depends on the event kinematics , and was therefore determined separately in two distinct regions of phase space .
the discriminating variable chosen for this purpose is @xmath148 , the highest transverse momentum of any track in the event . the average visible hadronic invariant mass , @xmath149 , and the resolution on @xmath146 as a function of the generated hadronic invariant mass @xmath147 for events with @xmath148 smaller and larger than 5 gev are shown in fig .
[ fig - wcorr](a ) and ( c ) .
the polynomial fits superimposed on fig .
[ fig - wcorr](b ) and ( d ) are used as a correction function so that @xmath150 .
the distributions are measured for invariant masses @xmath151 gev , @xmath152 gev , @xmath153 gev and @xmath154 gev where @xmath124 represents the hadronic invariant mass after correcting for detector effects . to facilitate comparisons to the results in @xcite , the ranges @xmath155 gev and @xmath156 gev
are also considered .
the following sources of systematic uncertainties have been investigated : * significant amounts of background need to be subtracted from the data at large @xmath108 , where the dominant source of background is multihadronic events ( @xmath127 ) .
the amount of this background subtracted from the data is varied by @xmath157% to estimate the uncertainty associated to this procedure .
the amount by which the background must be varied is estimated from studying the mc description of the data in regions of the phase space where the background dominates .
the effect on the measured cross - section is usually 1% or less for low transverse momenta , increasing to up to 7% at the highest momenta measured .
* the selection criteria described in section [ sec - evsel ] are varied simultaneously both to be more restrictive and to allow more events into the analysis to exclude a strong dependence on the event selection .
selection criteria based on energy measurements are varied by 5% in the ecal and hcal , and by 10% in the fd and sw calorimeters . the number of tracks required is changed by @xmath158 .
the allowed radial distance of the tracks is varied by 5% .
the uncertainty on the cross - section derived from all these variations is typically 1 - 6% . *
both phojet and pythia have been used to calculate the correction factors applied to the data .
the resulting distributions are averaged and the difference between the two distributions is used to define the systematic uncertainty .
the distributions obtained from both programs have been reweighted for a better description of the data .
the difference between using the reweighted and the original mc distributions to calculate the correction factors has been included in the systematic uncertainty .
the uncertainty derived from this study is below 5% . *
the systematic uncertainty due to the energy scale of the ecal for the range of energies in this analysis was estimated by varying the reconstructed ecal energy in the mc by @xmath159% @xcite .
the cross - sections change by up to 4% due to this variation . * studying vertex and net charge distributions it is estimated that about 2% of the selected events are due to beam - gas or beam - wall interactions .
this remaining background is treated as a systematic uncertainty of 2% .
systematic uncertainties due to the modelling of the detector resolution for the measurement of tracks @xcite and due to the luminosity measurement were found to be negligible .
the total systematic uncertainty was obtained by adding all systematic uncertainties in quadrature .
the differential inclusive cross - sections @xmath160 for charged hadrons in four ranges of the hadronic invariant mass @xmath124 are shown in fig .
[ fig - pt1 ] and are given in table [ tab - pt1a ] and table [ tab - pt1b ] .
the data points in the figure indicate the bin centre . both data and calculations are presented for quasi - real photons of virtualities @xmath161 gev@xmath106 , as described in sect . [ sect - kin ] .
calculations in nlo qcd @xcite are compared to the data .
the cross - sections are calculated using the qcd partonic cross - sections in nlo for direct , single- and double - resolved processes .
the hadronic cross - section is a convolution of the weizscker - williams effective photon distribution , the parton distribution functions and the fragmentation functions of @xcite .
the afg - ho parametrisation of the parton densities of the photon @xcite is used with @xmath162 mev .
the renormalisation and factorisation scales in the calculation are set equal to @xmath163 .
the cross - section calculation was repeated for the kinematic conditions of the present analysis . for the differential cross - section @xmath164 a minimum @xmath163 of 1.5 gev
is required to ensure the validity of the perturbative qcd calculation .
even at @xmath165 gev the cross - sections change by up to 80% when varying the renormalisation and factorisation scales by factors of two .
this uncertainty decreases rapidly to between 10% and 15% for @xmath166 gev and above .
the differential cross - section @xmath167 is hence restricted to the region @xmath168 gev to allow a meaningful comparison . in fig .
[ fig - pt1 ] the nlo calculation lies significantly below the data for transverse momenta greater than about 10 gev , which can be reached in the highest @xmath124 range only .
the predictions of phojet and pythia at high @xmath108 ( not shown ) are similar to the nlo calculation .
the differential cross - section is nearly independent of @xmath140 in the range measured as can be seen in fig .
[ fig - eta2 ] .
the data points in the figure indicate the bin centre .
the cross - section values are given in table [ tab - eta1 ] .
the nlo calculation reproduces the data well within the uncertainties of the calculation .
pythia and phojet describe the shape of the distributions correctly for all regions of the phase space measured , but are below the data in normalisation for large values of @xmath124 .
[ fig - pt2 ] ( a ) and ( b ) and table [ tab - ptl3 ] show the differential cross - section @xmath160 for charged hadrons for @xmath155 gev and @xmath156 gev to facilitate a comparison with a recent measurement by l3 of charged pions in the pseudorapidity range @xmath169 @xcite . the opal data shown in fig . [ fig - pt2 ] ( c ) and ( d ) and table [ tab - ptl3b ] have been scaled to account for the reduced @xmath140 range and for the fraction of charged pions of all charged hadrons using mc simulations . while the opal data points in fig .
[ fig - pt2 ] ( a ) and ( b ) are plotted at the bin centre , in fig .
[ fig - pt2 ] ( c ) and ( d ) they are plotted at the mean transverse momentum across the bin , as for the l3 data . from this comparison it is evident that the distributions measured by opal fall more rapidly towards high transverse momenta than those measured by l3 , leading to a disagreement between the two experiments at high transverse momenta and a better description of the opal data by nlo qcd than is the case for the l3 data .
we thank bernd kniehl for providing the nlo calculations and for many useful discussions .
we particularly wish to thank the sl division for the efficient operation of the lep accelerator at all energies and for their close cooperation with our experimental group .
in addition to the support staff at our own institutions we are pleased to acknowledge the + department of energy , usa , + national science foundation , usa , + particle physics and astronomy research council , uk , + natural sciences and engineering research council , canada , + israel science foundation , administered by the israel academy of science and humanities , + benoziyo center for high energy physics , + japanese ministry of education , culture , sports , science and technology ( mext ) and a grant under the mext international science research program , + japanese society for the promotion of science ( jsps ) , + german israeli bi - national science foundation ( gif ) , + bundesministerium fr bildung und forschung , germany , + national research council of canada , + hungarian foundation for scientific research , otka t-038240 , and t-042864 , + the nwo / nato fund for scientific research , the netherlands . + 99 l3 collaboration , p. achard et al . , phys .
b554 ( 2003 ) 105 .
opal collaboration , k. ackerstaff et al .
, z. phys .
c73 ( 1997 ) 433 ; + opal collaboration , g. abbiendi et al .
j. c10 ( 1999 ) 547 ; + opal collaboration , g. abbiendi et al .
j. c31 ( 2003 ) 307 .
l3 collaboration , p. achard et al .
b602 ( 2004 ) 157 .
the opal collaboration , k. ackerstaff et al .
j. c6 ( 1999 ) 253 .
opal collaboration , k. ahmet et al .
instrum .
methods a305 ( 1991 ) 275 ; + s. anderson et al .
instrum .
methods a403 ( 1998 ) 326 ; + opal collaboration , g. abbiendi et al .
j. c14 ( 2000 ) 373 .
t. sjstrand , comp .
commun . 82 ( 1994 ) 74 .
r. engel , z. phys .
c66 ( 1995 ) 203 ; + r. engel and j. ranft , phys .
d54 ( 1996 ) 4244 .
g.a . schuler and t. sjstrand , z. phys .
c68 ( 1995 ) 607 .
m. glck , e. reya and a. vogt , phys .
d45 ( 1992 ) 3986 ; + m. glck , e. reya and a. vogt , phys .
d46 ( 1992 ) 1973 .
s. jadach , b.f.l .
ward and z. was , comp .
commun . 79 ( 1994 ) 503 .
vermaseren , nucl .
b229 ( 1983 ) 347 .
g. marchesini , b.r .
webber , g. abbiendi , i.g .
knowles , m.h .
seymour and l. stanco , comp .
commun . 67 ( 1992 ) 465
. j. allison et al . , nucl .
instrum .
methods a317 ( 1992 ) 47 .
opal collaboration , g.abbiendi et al .
j. c14 ( 2000 ) 199 .
opal collaboration , k. ackerstaff et al .
j. c2 ( 1998 ) 213 ; + opal collaboration , g. abbiendi et al .
j. c12 ( 2000 ) 567 .
j. binnewies , b.a .
kniehl and g. kramer , phys .
d53 ( 1996 ) 6110 ; + b.a .
kniehl , private communications .
s. albino , b.a .
kniehl , g. kramer , nucl .
b725 ( 2005 ) 181 .
p. aurenche , j .-
guillet , m. fontannaz , z. phys .
c64 ( 1994 ) 621 .
.differential inclusive charged hadron production cross - sections @xmath164 for @xmath141 and in the @xmath124 ranges @xmath151 gev and @xmath152 gev .
the first uncertainty is the statistical uncertainty and the second uncertainty is the systematic uncertainty .
no value is given if the error on @xmath170 is less than 0.01 .
[ cols="^ , < , < , < , < " , ] | the inclusive production of charged hadrons in the collisions of quasi - real photons ( @xmath0+x ) has been measured using the opal detector at lep .
the data were taken at @xmath1 centre - of - mass energies from @xmath2 to @xmath3 gev .
the differential cross - sections as a function of the transverse momentum and the pseudorapidity of the hadrons are compared to theoretical calculations of up to next - to - leading order ( nlo ) in the strong coupling constant @xmath4 .
the data are also compared to a measurement by the l3 collaboration , in which a large deviation from the nlo predictions is observed .
european organization for nuclear research cern - ph - ep/2006 - 038 + 18th october 2006 + + + ( submitted to physics letters b ) the opal collaboration g.abbiendi@xmath5 , c.ainsley@xmath6 , p.f.kesson@xmath7 , g.alexander@xmath8 , g.anagnostou@xmath9 , k.j.anderson@xmath10 , s.asai@xmath11 , d.axen@xmath12 , i.bailey@xmath13 , e.barberio@xmath14 , t.barillari@xmath15 , r.j.barlow@xmath16 , r.j.batley@xmath6 , p.bechtle@xmath17 , t.behnke@xmath17 , k.w.bell@xmath18 , p.j.bell@xmath9 , g.bella@xmath8 , a.bellerive@xmath19 , g.benelli@xmath20 , s.bethke@xmath15 , o.biebel@xmath21 , o.boeriu@xmath22 , p.bock@xmath23 , m.boutemeur@xmath21 , s.braibant@xmath5 , r.m.brown@xmath18 , h.j.burckhart@xmath7 , s.campana@xmath20 , p.capiluppi@xmath5 , r.k.carnegie@xmath19 , a.a.carter@xmath24 , j.r.carter@xmath6 , c.y.chang@xmath25 , d.g.charlton@xmath9 , c.ciocca@xmath5 , a.csilling@xmath26 , m.cuffiani@xmath5 , s.dado@xmath27 , a.de roeck@xmath7 , e.a.de wolf@xmath28 , k.desch@xmath17 , b.dienes@xmath29 , j.dubbert@xmath21 , e.duchovni@xmath30 , g.duckeck@xmath21 , i.p.duerdoth@xmath16 , e.etzion@xmath8 , f.fabbri@xmath5 , p.ferrari@xmath7 , f.fiedler@xmath21 , i.fleck@xmath22 , m.ford@xmath16 , a.frey@xmath7 , p.gagnon@xmath31 , j.w.gary@xmath20 , c.geich-gimbel@xmath32 , g.giacomelli@xmath5 , p.giacomelli@xmath5 , m.giunta@xmath20 , j.goldberg@xmath27 , e.gross@xmath30 , j.grunhaus@xmath8 , m.gruw@xmath7 , a.gupta@xmath10 , c.hajdu@xmath26 , m.hamann@xmath17 , g.g.hanson@xmath20 , a.harel@xmath27 , m.hauschild@xmath7 , c.m.hawkes@xmath9 , r.hawkings@xmath7 , g.herten@xmath22 , r.d.heuer@xmath17 , j.c.hill@xmath6 , d.horvth@xmath33 , p.igo-kemenes@xmath23 , k.ishii@xmath11 , h.jeremie@xmath34 , p.jovanovic@xmath9 , t.r.junk@xmath35 , j.kanzaki@xmath36 , d.karlen@xmath13 , k.kawagoe@xmath11 , t.kawamoto@xmath11 , r.k.keeler@xmath13 , r.g.kellogg@xmath25 , b.w.kennedy@xmath18 , s.kluth@xmath15 , t.kobayashi@xmath11 , m.kobel@xmath37 , s.komamiya@xmath11 , t.krmer@xmath17 , a.krasznahorkayjr.@xmath38 , p.krieger@xmath39 , j.von krogh@xmath23 , t.kuhl@xmath40 , m.kupper@xmath30 , g.d.lafferty@xmath16 , h.landsman@xmath27 , d.lanske@xmath41 , d.lellouch@xmath30 , j.letts@xmath42 , l.levinson@xmath30 , j.lillich@xmath22 , s.l.lloyd@xmath24 , f.k.loebinger@xmath16 , j.lu@xmath43 , a.ludwig@xmath37 , j.ludwig@xmath22 , w.mader@xmath37 , s.marcellini@xmath5 , a.j.martin@xmath24 , t.mashimo@xmath11 , p.mttig@xmath44 , j.mckenna@xmath12 , r.a.mcpherson@xmath13 , f.meijers@xmath7 , w.menges@xmath17 , f.s.merritt@xmath10 , h.mes@xmath45 , n.meyer@xmath17 , a.michelini@xmath5 , s.mihara@xmath11 , g.mikenberg@xmath30 , d.j.miller@xmath46 , w.mohr@xmath22 , t.mori@xmath11 , a.mutter@xmath22 , k.nagai@xmath24 , i.nakamura@xmath47 , h.nanjo@xmath11 , h.a.neal@xmath48 , r.nisius@xmath15 , s.w.oneale@xmath49 , a.oh@xmath7 , m.j.oreglia@xmath10 , s.orito@xmath50 , c.pahl@xmath15 , g.psztor@xmath51 , j.r.pater@xmath16 , j.e.pilcher@xmath10 , j.pinfold@xmath52 , d.e.plane@xmath7 , o.pooth@xmath41 , m.przybycie@xmath53 , a.quadt@xmath15 , k.rabbertz@xmath54 , c.rembser@xmath7 , p.renkel@xmath30 , j.m.roney@xmath13 , a.m.rossi@xmath5 , y.rozen@xmath27 , k.runge@xmath22 , k.sachs@xmath19 , t.saeki@xmath11 , e.k.g.sarkisyan@xmath55 , a.d.schaile@xmath21 , o.schaile@xmath21 , p.scharff-hansen@xmath7 , j.schieck@xmath15 , t.schrner-sadenius@xmath56 , m.schrder@xmath7 , m.schumacher@xmath32 , r.seuster@xmath57 , t.g.shears@xmath58 , b.c.shen@xmath20 , p.sherwood@xmath46 , a.skuja@xmath25 , a.m.smith@xmath7 , r.sobie@xmath13 , s.sldner-rembold@xmath16 , f.spano@xmath59 , a.stahl@xmath41 , d.strom@xmath60 , r.strhmer@xmath21 , s.tarem@xmath27 , m.tasevsky@xmath61 , r.teuscher@xmath10 , m.a.thomson@xmath6 , e.torrence@xmath60 , d.toya@xmath11 , p.tran@xmath20 , i.trigger@xmath62 , z.trcsnyi@xmath38 , e.tsur@xmath8 , m.f.turner-watson@xmath9 , i.ueda@xmath11 , b.ujvri@xmath38 , c.f.vollmer@xmath21 , p.vannerem@xmath22 , r.vrtesi@xmath63 , m.verzocchi@xmath25 , h.voss@xmath64 , j.vossebeld@xmath65 , c.p.ward@xmath6 , d.r.ward@xmath6 , p.m.watkins@xmath9 , a.t.watson@xmath9 , n.k.watson@xmath9 , p.s.wells@xmath7 , t.wengler@xmath7 , n.wermes@xmath32 , g.w.wilson@xmath66 , j.a.wilson@xmath9 , g.wolf@xmath30 , t.r.wyatt@xmath16 , s.yamashita@xmath11 , d.zer-zion@xmath20 , l.zivkovic@xmath27 @xmath9school of physics and astronomy , university of birmingham , birmingham b15 2tt , uk @xmath5dipartimento di fisica dell universit di bologna and infn , i-40126 bologna , italy @xmath32physikalisches institut , universitt bonn , d-53115 bonn , germany @xmath20department of physics , university of california , riverside ca 92521 , usa @xmath6cavendish laboratory , cambridge cb3 0he , uk @xmath19ottawa - carleton institute for physics , department of physics , carleton university , ottawa , ontario k1s 5b6 , canada @xmath7cern , european organisation for nuclear research , ch-1211 geneva 23 , switzerland @xmath10enrico fermi institute and department of physics , university of chicago , chicago il 60637 , usa @xmath22fakultt fr physik , albert - ludwigs - universitt freiburg , d-79104 freiburg , germany @xmath23physikalisches institut , universitt heidelberg , d-69120 heidelberg , germany @xmath31indiana university , department of physics , bloomington in 47405 , usa @xmath24queen mary and westfield college , university of london , london e1 4ns , uk @xmath41technische hochschule aachen , iii physikalisches institut , sommerfeldstrasse 26 - 28 , d-52056 aachen , germany @xmath46university college london , london wc1e 6bt , uk @xmath16department of physics , schuster laboratory , the university , manchester m13 9pl , uk @xmath25department of physics , university of maryland , college park , md 20742 , usa @xmath34laboratoire de physique nuclaire , universit de montral , montral , qubec h3c 3j7 , canada @xmath60university of oregon , department of physics , eugene or 97403 , usa @xmath18cclrc rutherford appleton laboratory , chilton , didcot , oxfordshire ox11 0qx , uk @xmath27department of physics , technion - israel institute of technology , haifa 32000 , israel @xmath8department of physics and astronomy , tel aviv university , tel aviv 69978 , israel @xmath11international centre for elementary particle physics and department of physics , university of tokyo , tokyo 113 - 0033 , and kobe university , kobe 657 - 8501 , japan @xmath30particle physics department , weizmann institute of science , rehovot 76100 , israel @xmath17universitt
hamburg / desy , institut fr experimentalphysik , notkestrasse 85 , d-22607 hamburg , germany @xmath13university of victoria , department of physics , p o box 3055 , victoria bc v8w 3p6 , canada @xmath12university of british columbia , department of physics , vancouver bc v6 t 1z1 , canada @xmath52university of alberta , department of physics , edmonton ab t6 g 2j1 , canada @xmath26research institute for particle and nuclear physics , h-1525 budapest , p o box 49 , hungary @xmath29institute of nuclear research , h-4001 debrecen , p o box 51 , hungary @xmath21ludwig - maximilians - universitt mnchen , sektion physik , am coulombwall 1 , d-85748 garching , germany @xmath15max - planck - institute fr physik , fhringer ring 6 , d-80805 mnchen , germany @xmath48yale university , department of physics , new haven , ct 06520 , usa @xmath67 and at triumf , vancouver , canada v6 t 2a3 @xmath68 now at university of alberta @xmath69 and institute of nuclear research , debrecen , hungary @xmath70 now at institute of physics , academy of sciences of the czech republic 18221 prague , czech republic @xmath71 and department of experimental physics , university of debrecen , hungary @xmath72 and mpi mnchen @xmath73 and research institute for particle and nuclear physics , budapest , hungary @xmath74 now at university of liverpool , dept of physics , liverpool l69 3bx , u.k .
@xmath75 now at dept .
physics , university of illinois at urbana - champaign , u.s.a .
@xmath76 and manchester university manchester , m13 9pl , united kingdom @xmath77 now at university of kansas , dept of physics and astronomy , lawrence , ks 66045 , u.s.a .
@xmath78 now at university of toronto , dept of physics , toronto , canada @xmath44 current address bergische universitt , wuppertal , germany @xmath79 now at university of mining and metallurgy , cracow , poland @xmath42 now at university of california , san diego , u.s.a .
@xmath80 now at the university of melbourne , victoria , australia @xmath81 now at iphe universit de lausanne , ch-1015 lausanne , switzerland @xmath82 now at iekp universitt karlsruhe , germany @xmath83 now at university of antwerpen , physics department , b-2610 antwerpen , belgium ; supported by interuniversity attraction poles programme
belgian science policy @xmath84 now at technische universitt , dresden , germany @xmath85 and high energy accelerator research organisation ( kek ) , tsukuba , ibaraki , japan @xmath86 now at university of pennsylvania , philadelphia , pennsylvania , usa @xmath87 now at triumf , vancouver , canada @xmath88 now at desy zeuthen @xmath89 now at cern @xmath90 now at desy @xmath91 deceased |
since kauffman and lomonaco pointed out the relations between quantum entanglement and topological entanglement @xcite , the @xmath8 braid matrix and yang - baxter equation(ybe ) have been widely applied to entanglement and quantum information from various aspects @xcite .
for example , the solution of ybe was used for describing 2-qubit entanglement @xcite and 3-qubit entanglement @xcite , the entanglement was connected with berry phase in yang - baxter system @xcite , quantum optical realization of ybe @xcite , entanglement in describing quantum criticality of ybe chain model @xcite and so on . on the contrary , sometimes the entanglement can also be employed in obtaining braiding out of yang - baxter equation(ybe ) .
one of which is to detect the role of extremum of @xmath3-norm and von neumann entropy of qubit entangled state in determining the @xmath8 braid matrix from the parametrized @xmath2-matrix which satisfies yang - baxter equation @xcite . in previous works , the relations between quantum entanglement and localized unitary representation of braiding as well as @xmath7-matrix have been well discussed .
however , most of the results are limited in the qubit entanglement and @xmath8 braid matrix .
recently , based on the new localized @xmath9 representation of braid group associated with @xmath10 parafermion @xcite , the relations between qudit entanglement and braiding have been discussed @xcite . due to the potential applications of the @xmath11 parafermionic type of topological quantum models in quantum computing and quantum information @xcite , it is valuable to investigate further the relations between qudit entanglement and @xmath9 braiding as well as ybe
. now we briefly introduce some results about @xmath8 ybe in quantum information .
different from the original 6-vertex and 8-vertex @xmath8 localized unitary solutions of ybe originating from chain models , in n - qubit tensor product space @xmath12 , the new type of @xmath8 localized solution @xmath7-matrix associated with quantum information reads ( @xmath13 is @xmath14 identity matrix ) @xcite @xmath15\label{4solution}.\end{aligned}\ ] ] eq .
( [ ybecondition ] ) is the so - called yang - baxter equation . under the solution shown in eq .
( [ 4solution ] ) , the constraint for parameters in eq .
( [ ybecondition ] ) obeys the lorentzian type additivity @xmath16 , whereas the original 6-vertex solutions obey the galilean type additivity @xmath17 .
acting @xmath7-matrix on 2-qubit natural basis @xmath18 , where @xmath19 , one obtains four 2-qubit pure states . without loss of generality , we choose one
state @xmath20 clearly , the parameter @xmath21 in @xmath7 describes the entangled degree of @xmath22 .
when @xmath23 , @xmath22 turns to be the maximal entangled state , meanwhile , the @xmath7-matrix becomes braid matrix associated with @xmath24 chern - simons theory , @xmath25,\ ] ] i.e. the braid matrix describes the maximal entangled state . in comparison with braid @xmath26 ,
the advantage of @xmath2 is that the @xmath21 describes any entangled degree of 2-qubit pure states . on the other hand , from the viewpoint of @xmath3-norm
, the extremum of @xmath3-norm of @xmath22 exactly corresponds to the maximal entanglement as well as the braid matrix@xcite .
hence , @xmath3-norm endows the real physical meaning for @xmath2-matrix : the braiding operation . in this paper , we mainly focus on the role of @xmath3-norm and 2-qutrit entanglement playing in braiding representation associated with @xmath0 parafermions , which are related to metaplectic anyon models in @xmath27 chern - simons(cs ) theory @xcite .
firstly , based on the @xmath0 parafermion representation of yang - baxter equation , we investigate the application of @xmath3-norm and von neumann entropy of the 2-qutrit state generated by @xmath7-matrix , and find that the extreme values of @xmath3-norm and von neumann entropy correspond to the braiding .
the result is the generalization of qubit case proposed in ref .
@xcite . secondly , inspired by the qubit representation of 4-anyon basis for @xmath24 cs associated with @xmath8
braid matrix @xcite , the two 4-anyon topological fusion basis in @xmath27 cs are represented by 4-qutrit entangled states .
then the well known @xmath5 localized unitary representation of braiding is reduced to the jones representation of braiding for 4-strand topological basis .
the applications of @xmath3-norm in quantum physics have already been proposed in recent years , such as in quantum process tomography @xcite , yang - baxter equation @xcite and quantifying coherence @xcite , _ et al .
_ in the previous work @xcite , the authors have shown the motivation of adopting @xmath3-norm in real physical model associated with @xmath24 cs . here
we give a brief introduction about it .
usually in quantum mechanics , a wave function @xmath28 can be expanded as @xmath29 , where @xmath30 is orthonormal basis .
the normalization of @xmath28 reads @xmath31 we call @xmath32 as @xmath33-norm , which indicates the square integrability or the probability distribution of the wave function .
meanwhile , @xmath3-norm is defined as latexmath:[\[\label{l1norm } @xmath3-norm plays important roles in information theory such as compressed sensing theory @xcite , _ et al . _
hence , it is worthy of investigating whether the extremization of @xmath3-norm can be used to determine some important physical quantities in quantum information or in quantum mechanics . in ref .
@xcite , the extremum of @xmath3-norm has been well connected to @xmath8 braid matrix , we now extend the results to @xmath5 braid matrix .
the paper is organized as follows . in sec .
[ sec2 ] , we review the solutions of yang - baxter equation expressed by @xmath0 parafermion . in sec .
[ sec3 ] , the @xmath3-norm and von neumann entropy of the entangled state @xmath1 generated by @xmath7-matrix is discussed . in sec .
[ sec4 ] , the @xmath5 braid matrix are reduced to @xmath14 jones representation of braiding under the 4-strand topological basis represented by 4-qutrit entangled states . in the last section , we make the conclusion and discussion .
the metaplectic anyons have shown their universal topological quantum computation abilities under the braiding and measurement @xcite , hence it is valuable to study the characteristics of the type of anyons . in this section ,
we review the @xmath0 parafermionic representation @xcite of braid operators with quantum dimension @xmath35 , which is associated with metaplectic anyons in @xmath27 topological quantum field theory . based on the @xmath36 matrix representation of parafermions in tensor product space @xmath37 ,
the localized @xmath5 unitary representation for braid relation ( also known as yang - baxter equation without parameter ) can be obtained . moreover ,
the rational yang - baxterization @xcite of the corresponding braid operators are shown below .
we first introduce the @xmath0 parafermion . as a natural generalization of clifford algebra for majorana fermion ,
the algebraic relation of @xmath0 parafermion @xmath38 reads @xmath39 then the temperley - lieb(t - l ) elements associated with braid relation are composed by nearest neighbor parafermions @xmath40 which satisfies temperley - lieb algebra @xcite for quantum dimension @xmath35 , @xmath41 the braid operator is expressed as @xmath42 and satisfies braid relation @xcite @xmath43 by defining the concrete matrix representation of @xmath0 parafermion in n - qutrit space @xmath37 through ( @xmath13 is @xmath44 identity matrix ) @xcite @xmath45 where @xmath46 , \ , x=\left[\begin{array}{ccc } 0 & 1 & 0 \\ 0 & 0 & 1 \\ 1 & 0 & 0 \end{array}\right],\ ] ] one can obtain the braid matrix from three nearest parafermionic sites in 2-qutrit space @xmath47 : @xmath48\otimes\left[\begin{array}{ccc } 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{array}\right],\label{b12}\\ & & b_{23}=b_{2}=\frac{\textrm{i}\omega}{\sqrt{3}}\left[\begin{matrix } \omega & 0 & 0 & 0 & 0 & \omega & 0 & 1 & 0\\ 0 & \omega & 0 & \omega & 0 & 0 & 0 & 0 & 1\\ 0 & 0 & \omega & 0 & \omega & 0 & 1 & 0 & 0\\ 0 & 1 & 0 & \omega & 0 & 0 & 0 & 0 & \omega\\ 0 & 0 & 1 & 0 & \omega & 0 & \omega & 0 & 0\\ 1 & 0 & 0 & 0 & 0 & \omega & 0 & \omega & 0\\ 0 & 0 & \omega & 0 & 1 & 0 & \omega & 0 & 0\\ \omega & 0 & 0 & 0 & 0 & 1 & 0 & \omega & 0\\ 0 & \omega & 0 & 1 & 0 & 0 & 0 & 0 & \omega \end{matrix}\right],\label{b23}\end{aligned}\ ] ] obeying @xmath49 here we emphasize that the braid relation in eq.([b123 ] ) and ybe is expressed in terms of three nearest parafermions , which only occupy 2-qutrit space @xmath50 totally . under the rational yang - batxerization of braid matrix , the unitary @xmath7-matrix satisfying ybe shown in eq .
( [ ybecondition ] ) can be obtained @xcite @xmath51 which reduces braid matrix at @xmath4 and corresponds to the maximal von neumann entropy of qutrit states we shall show below . here
we note that the ybe in eq .
( [ ybecondition ] ) is independent of the concrete representation of @xmath2 , but the relation of three parameters is related to the @xmath7 . under the solution shown in eq .
( [ 9solution1 ] ) , the constraint from ybe leads to the condition for three parameters @xcite @xmath52 different from the traditional parameter relation @xmath17(galilean type additivity ) in yang - baxter equation in the usual chain models @xcite , the new relation for three parameters in eq .
( [ ybeangularrelation ] ) obeys lorentzian type additivity @xmath53 for @xmath54 .
in this section , we shall show that the maximal 2-qutrit von neumann entropy is determined by the braid matrix from parametrized @xmath2-matrix , i.e. @xmath4 for ybe .
the braid matrix can also be obtained by extremization of @xmath7-matrix .
we first obtain the reduced @xmath44 @xmath7-matrix as well as braid matrix from the original @xmath5 ones . under the 2-qutrit orthonormal basis @xmath55 , where @xmath56
, the @xmath7-matrix can be expressed by the ket - bra representation , as @xmath57 under the unitary transformation @xmath58 , @xmath59,\ ] ] the @xmath5 basis @xmath60 is reordered to be @xmath61 .
then it is not difficult to find that the two @xmath7 s are reducible and can be expressed as the direct sum of three @xmath44 matrices in three 3-d subspaces , @xmath62 , @xmath63 , @xmath64 , respectively .
indeed , the three subspaces can be categorized by the 2-qutrit `` parity operator '' defined as follows . introducing the 2-qutrit `` parity operator '' @xmath65 ,
@xmath66 it is easy to check that @xmath67 in the subspace @xmath62 , @xmath68 in the subspace @xmath63 , @xmath69 in the subspace @xmath64 , and @xmath7-matrix commutes with the parity @xmath65 , @xmath70=[\breve{r}_{23},p]=0.\ ] ] that is to say , the @xmath7 operation(including braid operation ) preserves the 2-qutrit parity @xmath65 of the system .
taking one subspace @xmath62 as example , the reduced @xmath44 @xmath7-matrix reads @xmath71 , \\ & & \mathcal{a}_{23}(\theta)=\left[\begin{smallmatrix}\cos\theta-\frac{i}{3}\sin\theta & \frac{2i}{3}\omega^2\sin\theta & \frac{2i}{3}\omega\sin\theta \\ \frac{2i}{3}\omega\sin\theta & \cos\theta-\frac{i}{3}\sin\theta & \frac{2i}{3}\omega^2\sin\theta \\ \frac{2i}{3}\omega^2\sin\theta & \frac{2i}{3}\omega\sin\theta & \cos\theta-\frac{i}{3}\sin\theta \end{smallmatrix}\right ] .
\end{aligned}\ ] ] they satisfy ybe @xmath72 with the same constraint as pointed in eq.([ybeangularrelation ] ) @xmath73 now we introduce how to screen out the braid relation from ybe by extremizing @xmath3-norm and von neumann entropy of the state generated by @xmath7-matrix . acting @xmath74 on basis @xmath75 ,
one obtains the state in the subspace @xmath62 @xmath76 following the definition in eq .
( [ l1norm ] ) , the @xmath3-norm of the state @xmath1 is expressed as @xmath77 on the other hand , the von neumann entropy @xcite , which reflects the quantum entangled degree of the two subsystems , reads @xmath78,\ ] ] where @xmath79 represents the partial trace of the density matrix of the system . in our case with the quantum state @xmath1 ,
the von neumann entropy is expressed by @xmath21 , as @xmath80 the values of @xmath3-norm and von neumann entropy @xmath81 , as functions of parameter @xmath21 in @xmath7-matrix , are shown in fig .
[ l1vn ] . from fig .
[ l1vn ] we conclude that the location of the extremum of @xmath3-norm and von neumann entropy of @xmath1 coincide each other exactly . especially , the two quantities both achieve the maximal values at @xmath82 , meanwhile the @xmath2 turns to be braid matrix .
the result is also applied to the qubit cases , which has been discussed in ref .
hence we conclude that the extreme values of the two quantities @xmath83 and @xmath81 endow the real physical meaning for @xmath7-matrix .
-norm of @xmath1 as a function of @xmath21 that describes the qutrit entanglement .
the von neumann entropy is labeled by the red dashed line , and the @xmath3-norm is labeled by blue solid line .
both of them achieve the maximal extremum at @xmath82 , at which value the @xmath2 reduces braid matrix . ]
in this section , we give an entangled 4-qutrit representation of 4-anyon topological basis . by acting the localized @xmath5 unitary representation of t - l elements on the topological basis , we obtain the @xmath14 representation of t - l algebra , which exactly corresponds to the the jones representation of the braid groups obtained from diagrammatic topological fusion basis . to start with , let us briefly review some basic knowledge about the graphic representation of braiding , temperley - lieb algebra and topological fusion basis@xcite
. graphically , the t - l algebra reads @xmath84{ununun}}}={\raisebox{-0.4\height}{\includegraphics[height=1.2cm]{unl } } } , \quad\ \
{ \raisebox{-0.4\height}{\includegraphics[height=1.2cm]{unun}}}={\raisebox{-0.4\height}{\includegraphics[height=1.2cm]{uon}}}={\raisebox{-0.4\height}{\includegraphics[height=0.4cm]{loop}}}\ { \raisebox{-0.52\height}{\includegraphics[height=0.8cm]{un}}},\ ] ] @xmath85{unij}}},\quad d={\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{loop}}}\ ( \text{value of loop}).\ ] ] the braid operation @xmath86{cross}}}$ ] in eq .
( [ btl ] ) can be decomposed into the combination of identity operator and t - l element under the skein relation , graphically , as @xmath87{cross}}}=-e^{\frac{i\pi}{12}}\left(\alpha{\raisebox{-0.52\height}{\includegraphics[height=0.8cm]{ll}}}+\alpha^{-1}{\raisebox{-0.52\height}{\includegraphics[height=0.8cm]{un}}}\right ) , \ , d={\raisebox{-0.4\height}{\includegraphics[height=0.5cm]{loop}}},\ ] ] where @xmath88 , and quantum dimension @xmath89 . in @xmath27 chern - simons(cs )
theory , the dimension of hilbert space for 4-anyon fusion states is two @xcite , with each anyon has topological charge 1 . under the jones - wenzl idempotent @xcite , the two orthonormal fusion basis can be expressed as cups @xmath90{uu}}},\\ & |e_2\rangle = \frac{1}{\sqrt{d^2 - 1}}\left ( { \raisebox{-0.4\height}{\includegraphics[height=0.6cm]{twojoin}}}-\frac{1}{d } { \raisebox{-0.4\height}{\includegraphics[height=0.6cm]{uu}}}\right ) .
\end{split}\ ] ] acting the t - l elements on the above two basis @xmath91 , one obtains @xmath92 , \\ & t_{23}=t_2=\frac{1}{\sqrt{3}}\left[\begin{array}{cc } 1 & \sqrt{2 } \\
\sqrt{2 } & 2 \end{array}\right ] , \\ &
t_{34}=t_3=\left[\begin{array}{cc } \sqrt{3 } & 0 \\ 0 & 0 \end{array}\right ] , \end{split}\ ] ] which correspond to the jones representation of the braid group @xcite . in the previous paper @xcite , the @xmath8 localized unitary representation of braid operator with quantum dimension @xmath93 has been connected with the jones representation of braid under the qubit realization of 4-anyon fusion basis for @xmath24 chern - simons theory .
we now extend the result to @xmath5 metaplectic anyon system with quantum dimension @xmath35 associated with the @xmath27 . due to the simple relation between braid operators and t - l elements as shown in eq .
( [ btl ] ) , we consider the t - l elements directly .
the @xmath94 localized unitary representation of t - l algebra in @xmath37 reads ( @xmath13 is @xmath44 identity matrix ) @xcite @xmath95 . \end{split}\ ] ] under the representation of @xmath0 parafermions shown in eq .
( [ z3jw ] ) , @xmath96 can be reexpressed by eq .
( [ z3para ] ) , as @xmath97 since the braid matrix and t - l matrix are closely related to the 2-qutrit maximal entangled states defined by maximal von neumann entropy , it is reasonable to express the t - l elements by 2-qutrit entangled states with maximal entangled states .
considering the 2-qutrit parity symmetry of the t - l matrix , we introduce three maximal entangled states , ( @xmath98 ) @xmath99 where @xmath100 represent 2 different qutrit sites .
it is interesting that the t - l elements @xmath96 in eq .
( [ lurt ] ) can be expressed by the three maximal entangled states , as ( j = i+1 ) @xmath101 moreover , the 4-anyon topological basis in eq.([topological basis ] ) for @xmath35 can also be represented by 4-qutrit states , @xmath102 in the 4-qutrit space @xmath103 , there are totally three t - l elements .
acting @xmath96 in eq . ( [ tlp ] ) on the two basis in eq .
( [ tbqutrit ] ) , one obtains @xmath104,\\ & t'_{23}=e'_2=\frac{1}{\sqrt{3}}\left[\begin{array}{cc } 1 & \sqrt{2 } \\
\sqrt{2 } & 2 \end{array}\right],\\ & t'_{34}=e'_3=\left[\begin{array}{cc } \sqrt{3 } & 0 \\ 0 & 0 \end{array}\right ] .
\end{split}\ ] ] substituting eq.([qtl2 ] ) into eq.([btl ] ) , we obtain the @xmath14 braid matrices @xmath105,\\ & b'_{23}=b'_2=\frac{1}{\sqrt{3}}\left[\begin{array}{cc } e^{-i\frac{\pi}{6 } } & i\sqrt{2 } \\
i\sqrt{2 } & e^{i\frac{\pi}{6 } } \end{array}\right],\\ & b'_{34}=b'_3=\left[\begin{array}{cc } e^{i\frac{\pi}{3 } } & 0 \\ 0 & e^{-i\frac{\pi}{3 } } \end{array}\right ] .
\end{split}\ ] ] in comparison with eq .
( [ gtl2 ] ) , we find that the obtained @xmath14 @xmath106 are exactly the jones representation of braid matrix in 4-strand space .
thus by constructing the parity - preserved qutrit entangled states with maximal von neumann entropy and expressing the t - l elements and topological basis by the entangled states , we obtain the relationship between localized unitary representation and jones representation of braid for 4-strand system .
this paper mainly includes two results about @xmath0 parafermion representation of braiding associated with quantum information .
all of the results are related to reducing higher dimensional representation of braid matrices to the lower ones .
the first is reducing the @xmath5 braid matrices in eq .
( [ b12 ] ) and eq .
( [ b23 ] ) to @xmath44 by introducing parity @xmath65-preserved subspace , such as @xmath62 ( one can also choose other two subspaces , which do not change the results ) . in the subspace , the braid operators with real physical meaning can be screened out from @xmath7-matrix with the aid of 2-qutrit maximal entangled states , where both @xmath3-norm and von neumann entropy achieve their maximal values .
the second is reducing the localized @xmath5 unitary representation of braiding relating to t - l elements in eq .
( [ lurt ] ) into @xmath14 jones representation in 4-anyon basis . under the representation ,
the behavior of localized braid matrix coincides well with the jones representation of braiding , which can be obtained by braiding the fusion basis diagrammatically . in this sense
, the qutrit representation does reflect the braiding properties .
however , how to propose a well defined topological entanglement entropy for fusion basis , as proposed in ref .
@xcite , by means of qutrit language is still in challenge .
hence , the result can only be regarded as a qutrit simulation of diagrammatic braiding operation for 4-strand system , where each strand corresponds to a sector with topological charge 1 . indeed , both of the above two results rely partly on the maximal entangled states(defined by von neumann entropy )
. these are due to the parity preservation ( see eq .
( [ parity ] ) ) of the braiding operation on the natural basis @xmath107 . concretely , for 2-qutrit pure states , the @xmath5 hilbert space can be classified into three @xmath44 subspaces by eigenvalues of parity @xmath65 . taking one subspace @xmath62 as example , where @xmath67 , the superposed state @xmath108 achieves its maximal von neumann entropy and @xmath3-norm simultaneously at equal distribution @xmath109 .
it not only reflects the role of parity @xmath65 plays in generating entanglement in the 2-qutrit hilbert space , but also guarantees the @xmath3-norm and von neumann entropy achieves their maximal values simultaneously . some of the results can also be extended to @xmath110 parafermion solution of ybe , but for @xmath111 , the rational yang - baxterized @xmath7-matrix is not unitary . in conclusion , we take @xmath0 parafermion model as examples to show the important roles of the parity , @xmath3-norm and von neumann entropy for 2-qutrit in determining the braiding behavior of yang - baxter equation . how to make connections between n - qudit entanglement and n - strand localized unitary @xmath9 representation of braid group is still an open problem . on the other hand , extending the 4-qutrit realization of 4-anyon topological basis to n - anyon basis is also worthy of doing .
this work is in part supported by nsf of china(grants no .
11475088 ) . | * the relationships between quantum entangled states and braid matrices have been well studied in recent years .
however , most of the results are based on qubits . in this paper
, we investigate the applications of 2-qutrit entanglement in the braiding associated with @xmath0 parafermion .
the 2-qutrit entangled state @xmath1 , generated by acting the localized unitary solution @xmath2 of ybe on 2-qutrit natural basis , achieves its maximal @xmath3-norm and maximal von neumann entropy simultaneously at @xmath4 .
meanwhile , at @xmath4 , the solutions of ybe reduces braid matrices , which implies the role of @xmath3-norm and entropy plays in determining real physical quantities . on the other hand
, we give a new realization of 4-anyon topological basis by qutrit entangled states , then the @xmath5 localized braid representation in 4-qutrit tensor product space @xmath6 are reduced to jones representation of braiding in the 4-anyon topological basis .
hence , we conclude that the entangled states are powerful tools in analysing the characteristics of braiding and @xmath7-matrix . * |
our main tool of analysis is a result attributed to jordan @xcite , a more modern version of which can be found in @xcite . the result states that any two projectors @xmath69 and @xmath18 on an @xmath0-dimensional hilbert space can be put in a simultaneous block - diagonal form , with blocks of size at most 2 .
clearly , the operators @xmath72 and @xmath83 share the same spectrum @xmath84 , bounded between 0 and 1 ( this can be seen by left and right polar decomposition of the operator @xmath85 ) .
their respective eigenvectors @xmath86 and @xmath87 are in general different however . since @xmath69 and @xmath18 are purely contractive , it follows that if @xmath88 for some @xmath89 , then the basis can be chosen such that @xmath90 .
when @xmath91 , then either @xmath92 or @xmath93 . for the intermediate values @xmath94
, jordan found that we can group the remaining eigenvectors in pairs that we denote @xmath95 and @xmath96 such that @xmath97 with the property that @xmath98 and @xmath99 . by enlarging the dimension of the hilbert space ( at most doubling its dimension ) and allowing @xmath100 to take the values 0 and 1
, we can assume that all eigenstates of @xmath72 and @xmath83 are paired up in this fashion , so that both sets @xmath101 and @xmath102 are complete orthonormal bases .
the naive algorithm to approximate the ground state of a local hamiltonian consists of using phase estimation to mark the low - energy eigenstates of @xmath2 , and then integrating this marking procedure in grover s algorithm to produce a state of low energy .
this strategy can easily be analyzed with the help of jordan s result .
indeed , let @xmath69 be the projector onto the all - zero state of the @xmath29 auxiliary qubits used by phase estimation and @xmath18 be the projector on the output of the phase estimation indicating an energy less than @xmath15 .
as mentioned in the main text , the unitary evolution @xmath103 can not be realized perfectly in general , but this is a source of error that can be ignored because it does nt build up during grover s algorithm .
thus , we will assume here that the phase estimation algorithm implements @xmath104 exactly .
in this language , the state @xmath105 corresponds to a state of the form @xmath106 where @xmath30 is an eigenstate of @xmath2 with energy @xmath33 .
the eigenvalue @xmath73 of @xmath72 associated to that state is equal to the probability that the phase estimation algorithm diagnoses @xmath30 as having an energy less than @xmath15 .
this is where the finite success probability of the phase estimation algorithm comes into play .
the important point is that even if the energy @xmath33 is well below or well above @xmath15 , there is a small probability that the algorithm will give the wrong answer .
the effect is more pronounced when the actual energy of @xmath33 is close to the acceptance threshold @xmath15 , but it is the small error probabilities that will really cause a problem .
the error is zero only when @xmath33 can be expressed exactly in binary form with @xmath29 bits . thus , unless very detailed knowledge of the spectrum of @xmath2 is available , these small errors are unavoidable .
continuing with the naive algorithm , we prepare a random superposition @xmath107 , i.e. a random state of the @xmath12 system qubits with the @xmath29 auxiliary qubits in state 0 , and use grover s algorithm to amplify the image of @xmath18 .
this results in the state @xmath108 we see that , as a consequence of the amplification of @xmath18 , the state @xmath109 is no longer supported only on the image of @xmath69 as indicated by the presence of the @xmath110 in the state .
the overlap with @xmath69 is easily computed @xmath111 and this can be arbitrarily small .
for instance , if the vast majority of states have @xmath112 and a negligible fraction of them have @xmath73 s of order unity , this overlap is of order @xmath113 .
this example summarizes the main problem of the naive approach that is overcomed by our method .
we want to create a state that has a large overlap with two projectors @xmath69 and @xmath18 .
when @xmath71=0 $ ] , we can simply prepare a random state in the image of @xmath69 and use grover s algorithm to amplify @xmath18 .
but when the two projectors do not commute , the amplification of @xmath18 can move the state almost completely outside the image of @xmath69 , so the technique fails .
we make use of an upper and a lower bound on the inner - product of two different momentum states ( c.f .
. [ eq : momentum ] ) used by our algorithm .
1 ) @xmath114 and 2 ) when @xmath115 , then @xmath116 .
to prove the first bound , we use the inequality @xmath117 for @xmath118 and elementary algebra to obtain @xmath119 as claimed . for the second bound ,
assume that @xmath115 .
we can use the fact that @xmath120 to obtain @xmath121 raising to the @xmath43th power and using the inequality @xmath122 for @xmath123 and @xmath124 gives the desired result .
we will now demonstrate that our algorithm will work as advertised , i.e. produce a state of energy @xmath125 , with high probability .
we break this proof into two parts .
first , we show that the algorithm does not systematically abort and second that when it does not abort it produces the right state .
remember that the algorithm aborts when the overlap @xmath126 is less than @xmath127 . using the lower bound derived above ,
we know that if there is at least one eigenvalue , say @xmath128 , that lies in the interval @xmath129 , this overlap will be at least @xmath130 .
since the state is chosen at random , the amplitude @xmath131 has an exponentially high probability of being greater than @xmath132 which is sufficient for the success of the amplification procedure .
note that the interval @xmath129 has a polynomial width , i.e. @xmath29 scales only logarithmically with @xmath12 and @xmath43 polynomially .
thus , we can sweep all values of @xmath42 to this accuracy in polynomial time .
we must now demonstrate that when the amplification procedure succeeds , then with very high probability it generates a state with the desired energy .
we can assume in this case that @xmath133 since otherwise the amplification procedure would have had an exponentially small chance of succeeding .
let us first upper bound the average energy : @xmath134 where @xmath135 is the sum restricted to values of @xmath89 such that @xmath136 . using the lower bound on @xmath137 and the fact that @xmath138 we obtain @xmath139 consequently , the desired bound is obtained when @xmath140 .
using our upper bound on the inner product @xmath141 yields @xmath142 which is satisfied with @xmath143 and @xmath144 .
similar arguments produce an upper bound on the average energy @xmath145 where @xmath135 is now the sum restricted to values of @xmath89 such that @xmath146 .
using the lower bound on @xmath137 and the upper bound on the inner product @xmath39 gives @xmath147 which is greater than @xmath148 provided that @xmath149 .
note that with a @xmath29 growing polynomially with @xmath150 and @xmath43 growing polynomially with @xmath12 , the error on the energy decreases as @xmath34 and the total running time is @xmath151 as desired .
the analysis of the general algorithm for problems in qma follows similar lines , but the derivation of the filter function is slightly more involved .
the verification circuit @xmath6 acts on @xmath152 qubits : the @xmath12 qubits containing the witness and a @xmath67-qubit scratchpad .
we make use of jordan s result with @xmath69 being the projector onto the all - zero state of the scratchpad and @xmath153 is the projector onto the accepting subspace of the verification procedure .
remember that the verification procedure takes inputs of the form @xmath154 and accepts with probability @xmath155 .
thus , the eigenstates @xmath156 of @xmath72 correspond to witnesses with all - zero scratchpads with an accepting probability @xmath73 .
our algorithm makes use of @xmath29 additional auxiliary qubits , for a total of @xmath157 .
armed eqs ( [ jordan1]-[jordan4 ] ) , we see that the circuit of figure [ fig : circuit ] applied to the state @xmath158 produces the outcome @xmath159 indeed , we get factor of @xmath74 each time the outcome switches from 0 to 1 or vice versa , and a factor of @xmath160 otherwise . there is an additional factor of @xmath161 for every pair of consecutive 0 s .
let us now analyse our algorithm .
we initialize the first @xmath51 qubits in a random state @xmath19 in the support of @xmath18 .
this is done by setting the first qubit in the state @xmath162 and the remaining @xmath163 qubits in a random stabilizer state , and applying the unitary transformation @xmath164 .
then , we prepare the @xmath29 auxiliary qubits in the state @xmath75 . this can be achieved by a circuit composed of @xmath29 2-qubit gates .
we now apply the inverse of the circuit shown on figure [ fig : circuit ] , producing the state @xmath165 where the ellipsis represents terms outside the image of @xmath166 .
the filter function is defined by @xmath167 in the last step , we have approximated the binomial distributions by normal distributions . at this point , we can use grover s algorithm to amplify the image of @xmath166 , and the rest of the analysis proceeds as above . | preparing the ground state of a system of interacting classical particles is an np - hard problem .
thus , there is in general no better algorithm to solve this problem than exhaustively going through all @xmath0 configurations of the system to determine the one with lowest energy , requiring a running time proportional to @xmath0 .
a quantum computer , if it could be built , could solve this problem in time @xmath1 . here , we present a powerful extension of this result to the case of interacting _ quantum _ particles , demonstrating that a quantum computer can prepare the ground state of a quantum system as efficiently as it does for classical systems .
the simulation of quantum many - body systems is a notoriously hard problem in condensed matter physics , but it could easily be handled by a quantum computer @xcite .
there is however one catch : while a quantum computer can naturally implement the dynamics of a quantum system i.e. solve schrdinger s equation there was until now no general method to initialize the computer in a physically relevant state of the simulated system . for most physical applications , we are interested in the low - energy eigenstates of the hamiltonian @xmath2 because they describe the most interesting phases of matter , e.g. ferromagnetism , superconductivity , quantum hall effect , and bose - einstein condensation to name a few . unfortunately
, preparing low - energy states is already a very difficult task even when @xmath2 describes a classical system .
indeed , this problem is an archetype of the complexity class np .
this class contains all decision problems , i.e. problems of the form
does @xmath3 satisfy the property @xmath4 ? " , such that when the answer is yes " , there exists a witness @xmath5 that can be used to prove this answer efficiently .
more precisely , for each @xmath3 there exists a polynomial - size verification circuits @xmath6 such that 1 ) when @xmath7 there exists a witness @xmath5 that will cause @xmath6 to output @xmath8 , and 2 ) when @xmath9 , all witnesses cause @xmath6 to output @xmath10 .
consider for instance a local ising model @xmath11 on @xmath12 spins @xmath13 .
is there a spin configuration @xmath14 of energy less than @xmath15 ? " is a problem in np .
indeed , when the answer is yes , the configuration @xmath14 that achieves this low energy can serve as a witness . verifying the answer boils down to computing the energy , which requires at most @xmath16 operations .
finally , when the answer is no " , there is no configuration that can cause the verification procedure to accept .
clearly , an algorithm that solves this problem can be used to determine the ground state energy with little overhead .
the ising problem is in fact np - complete , meaning that it is the hardest problem in the class @xcite .
even more surprising is the fact that approximating the energy of the system with an error that increases with the system size @xmath12 is just as hard as the exact case it is also np - complete .
this is a consequence of a famous theorem on probabilistically checkable proofs ( pcp ) @xcite .
although some special cases can be solved efficiently @xcite , there is in general no better algorithm to solve the ising problem than systematically going through all @xmath17 spin configurations to determine the one with lowest energy . additionally to the immediate physical context
, finding ground states provides a very natural setting for studying combinatorial optimization problems problems that consist in minimizing an objective function @xmath2 ( playing the role of energy ) over some configuration space .
optimization problems play a vital role in almost every branch of science , from computer science to statistical physics and computational biology @xcite . determining a solution by exhaustive
search is , in general , computationally prohibitive because the size @xmath0 of the search space grows exponentially with the input size .
given the practial importance of optimization problems , more efficient methods are highly desirable .
a common strategy to solve optimization problems is simulated annealing @xcite .
like its name suggests , this method imitates the process undergone by a metal that is heated to a high temperature and then slowly cooled to its configuration of lowest energy .
if the cooling process is too fast , the system can become trapped in a local minimum , resulting in a failure of the algorithm . when the cooling is sufficiently slow however
, thermal fluctuations should prevent this phenomenon from occurring .
thus , simulated annealing requires a detailed knowledge of the energy landscape and therefore can not be applied to all minimization problems .
it was shown recently that a simulated annealing algorithm operated on a quantum computer achieves a quadratic speed - up over classical annealing @xcite .
whether the method can minimize the energy of a quantum system as efficiently is unknown .
adiabatic quantum computation is another method to tackle this class of problems with a quantum computer .
the adiabatic theorem asserts that a system prepared in the instantaneous ground state of a hamiltonian that varies slowly in time will remain in the ground state .
the ground state of @xmath2 can thus be prepared by choosing a time - dependent hamiltonian with a simple initial ground state and slowly changing it to @xmath2 .
this algorithm was applied to randomly generated instances of an np - complete problem @xcite .
the algorithm worked well for the small examples that could be simulated on a classical computer .
it was later shown @xcite however that the particular interpolation scheme suggested @xcite fails for satisfiability problems , and the best known upper bound is a running time of @xmath0 @xcite ( poly - logarithmic corrections are ignored throughout ) . in principle , this technique can also be applied to minimize the energy of a quantum hamiltonian , but little is known about its performances in that case .
finally , grover s algorithm @xcite can find the ground state of a classical system in @xmath1 steps .
given a projector @xmath18 and a state @xmath19 with @xmath20 , grover s algorithm consists of a sequence of two reflections , @xmath21 and @xmath22 .
repeating this sequence @xmath23 times has the effect of projecting @xmath19 onto the image of @xmath18 , with small corrections that can safely be ignored for the present discussion .
note that the value of @xmath24 must be approximately known , and this can be achieved by quantum counting @xcite .
choosing @xmath18 to be the projector on @xmath25 and @xmath19 a uniform superposition of all spin configurations yields , after at most @xmath1 iterations , a state of energy less than @xmath15 .
the ground state is obtained by sweeping " the value of @xmath15 .
although this remains an exponential scaling , it is significantly faster than a brute force search , and there are indications that this scaling is optimal @xcite . at first sight
, it seems like this last technique could be used to find the ground state of a quantum many - body system just as well .
all that is needed is a method to implement a projector @xmath18 onto the low - energy states of the system , i.e. @xmath26 for some given @xmath15 .
combining this method with grover s algorithm on an initial random state would create the desired outcome with high probability .
in fact , it is not necessary to initialize the system in a truly random state , but instead it can be randomly selected among all _ stabilizer _ states .
these have all the essential properties of random states and , most importantly , can be prepared with at most @xmath16 operations @xcite .
unfortunately , there is no known procedure to implement the projector on @xmath26 . the phase estimation algorithm
@xcite comes close however : it can be used to estimate the energy of any local hamiltonian with a polynomial small error and failure probability . to describe this algorithm
, it is convenient to assume that @xmath2 has been normalized such that @xmath27 and to consider its spectral decomposition , @xmath28 .
the phase estimation algorithm uses @xmath29 auxiliary qubits initially in the state 0 and , given an eigenstate @xmath30 of @xmath2 , produces the output @xmath31 where @xmath32 these are momentum " states , so the value of @xmath33 can be estimated via fourier transform .
hence , we can implement an approximation @xmath18 of the projector on @xmath26 by running the phase estimation algorithm and projecting the auxiliary qubits onto the subspace of low momentum . combining this method with grover s algorithm
should thus yield a good approximation of the ground state . however , a detailed analysis ( see appendix [ naive ] ) of this naive " approach reveals a failure .
the problem is that the projector @xmath18 constructed from phase estimation is only an approximation of @xmath26 and errors can build up during the amplification procedure .
there are two sources of errors .
firstly , the quantum computer can not exactly reproduce the dynamics of the many - body system .
this is not a problem however since a @xmath34 accuracy can be achieved using a trotter - suzuki decomposition at a polynomial cost @xcite , and this error does not build up .
we will henceforth safely ignore this source of error .
secondly , even when the energy @xmath33 associated with @xmath30 is well above the acceptance threshold @xmath15 , there is a small probability that phase estimation will diagnose it as being smaller than @xmath15 .
it is these imperfections that cause the algorithm to fail because they build up during amplification .
the typical outcome of this naive algorithm is an entangled state of the system and auxiliary qubits rather than a low - energy state of the system qubits tensored with the all - zero state of the auxiliary qubits .
detailed knowledge of the energy landscape such as the presence of an energy gap could be used to circumvent this effect , but in general the method will fail .
we will now present our algorithm that works for all local hamiltonians .
we proceed by making two modifications to the naive algorithm .
a detailed analysis is presented in the appendices .
the first modification is to run the algorithm backward : we initialize the system qubits in a random state @xmath35 , the auxiliary qubits in a low momentum state @xmath36 ( c.f .
eq .
[ eq : momentum ] ) , and execute the inverse of the phase estimation algorithm .
this produces the state @xmath37 where the ellipsis represents terms where the auxiliary qubits are in a state orthogonal to @xmath38 .
the factor @xmath39 is a function of @xmath40 peaked at 0 with a width @xmath41 .
thus , we can use grover s algorithm to amplify the all - zero state of the auxiliary qubits and obtain a state that is mostly a superposition of those eigenstates of @xmath2 with eigenvalues close to @xmath42 , i.e. the amplitude of each term in the superposition gets re - weighted by @xmath39 .
this procedure truly acts as a filter , suppressing the amplitude of eigenstates outside its bandwidth for benefit of the eigenstates inside the bandwidth .
moreover , the auxiliary qubits are systematically returned to @xmath10 as desired .
unfortunately , this is still not sufficient for our purpose because the filter has a heavy tail .
there is an exponential number of states with energy outside the bandwidth , so unless their amplitude is exponentially suppressed , they can significantly shift the energy of the state .
the filter we have constructed offers a polynomial suppression ; we need a filter that drops more abruptly outside its bandwidth .
this requires a second modification to the naive algorithm and is realized by repeating the phase estimation @xmath43 times , using a total of @xmath44 auxiliary qubits .
we obtain the same state @xmath45 as above ( c.f .
. [ eq : phi ] ) , except that the factor @xmath46 is now raised to the @xmath43th power . for those @xmath33 that are within @xmath47 of @xmath42 , this factor is at least @xmath48 .
thus , the overlap of this state with the projector @xmath49 will typically be @xmath50 where @xmath51 is the number of eigenstates of @xmath2 in the bandwidth of the filter .
thus , grover s algorithm can be used to amplify this overlap to nearly 1 in a time at most @xmath52 .
when the overlap of the state with the filter is too small , i.e. if @xmath53 say , this step will fail and the algorithm will abort . choosing @xmath54 and @xmath55 yields , after a successful application of grover s amplification , a state of energy @xmath56 as desired ( see appendix [ ground ] ) . to summarize , this algorithm acts as a filter on the energy . the position @xmath42 and width @xmath57 of the filter are specified by the user . when no eigenstates of @xmath2 lie within the filter s bandwidth , the algorithm aborts as desired .
note that the method can be adapted in a straightforward way to produce thermal distributions of the system at any temperature @xmath58 .
we could in a first step combine our method with quantum counting @xcite to estimate the density of states @xmath59 with a @xmath34 resolution .
we could then choose an energy scale @xmath15 at random according to the distribution @xmath60 and use our algorithm to prepare a state of energy close to @xmath15 .
analogously to the classical case , determining the ground state energy of a local quantum many - body system within accuracy @xmath34 is a complete problem for the complexity class known as quantum merlin and artur ( qma ) @xcite .
whether the problem remains complete when an extensive error is tolerated is unknown , but would be a natural quantum extension of the pcp theorem .
indeed , qma is a natural generalization of np : it is defined similarly except that both the witness and the verification circuit @xmath6 are quantum mechanical . beside the @xmath12 witness qubits on which it operates
, @xmath6 can also make use of @xmath61 auxiliary qubits initialized in the state @xmath10 that serve as a scratchpad during the computation .
the output of the verification procedure is obtained by measuring the first output qubit of the circuit .
because of the intrinsic randomness of quantum mechanics , this procedure is probabilistic : 1 ) when @xmath7 , there exists a witness @xmath5 that will cause @xmath6 to output @xmath8 with probability greater than @xmath62 , and 2 ) when @xmath9 , all witnesses cause @xmath6 to output @xmath8 with probability less than @xmath63 where @xmath64 .
the completeness of the local hamiltonian problem for the class qma suggests that our algorithm can be used to solve all these problems and prepare the relevant witness in a time @xmath65 .
this is not quite right because the mapping to the local hamiltonian problem does not preserve the size of the witness .
nevertheless , a small modification almost does the trick .
a good witness is a @xmath66 qubit state @xmath19 with 1 ) all @xmath67 ancillary qubits in the state @xmath10 summarized by @xmath68 where @xmath69 is the projector onto the all - zero state of the auxiliary qubits and 2 ) a probability at least @xmath63 of outputting @xmath8 at the verification procedure summarized by @xmath70 where @xmath18 is the projector associated with the verification procedure .
thus , the tasks of preparing a good witness boils down to producing a state that maximizes the overlap with two projectors @xmath69 and @xmath18 . when @xmath71 = 0 $ ] , this task can be accomplished straightforwardly with grover s algorithm , but additional efforts are required in the general case . indeed ,
if we start say with a random state in the image of @xmath69 and amplify the projector @xmath18 , we will typically obtain a state that is mostly supported outside the image of @xmath69 , unless @xmath71=0 $ ] . the method we propose to solve this problem is a modification of our algorithm that builds on the work of mariott and watrous @xcite .
the main modification is to replace the phase estimation algorithm by the circuit of figure [ fig : circuit ] .
it consists of a sequence of @xmath29 ( odd ) alternating measurements of @xmath18 and @xmath69 whose outcomes are coherently recorded on @xmath29 auxiliary qubits initially in the state 0 .
the behavior of this circuit on an eigenstate @xmath30 of the operator @xmath72 with eigenvalue @xmath73 can be analyzed using a result of jordan @xcite ( see appendix [ jordan ] ) .
the state of the @xmath29 auxiliary qubits becomes a superposition of all sequences of @xmath10 and @xmath8 , and the amplitude of consecutive distinct outcomes , i.e. the amplitude associated to each switches " from 0 to 1 or vice versa , is @xmath74 .
thus , counting the number of switches in the measurement outcomes allows us to estimate the eigenvalue @xmath73 of the state . and @xmath69 .
the results are coherently imprinted on @xmath29 auxiliary qubits . ]
the situation is therefore analogous to phase estimation , except that the eigenvalue @xmath73 is not encoded in a momentum state but in a state with a certain number of switches between the outcomes 0 and 1 .
accordingly , we must replace the momentum state used in our algorithm by a state with the right distribution of switches @xmath75 where @xmath76 denotes the number of switches and @xmath77 the number of pairs of consecutive @xmath10 s .
the bandwidth @xmath78 is adjusted by setting @xmath79 .
one important advantage of this type of filter state is that it drops very abruptly outside its bandwidth , @xmath80 is essentially proportional to a normal distribution centered at @xmath81 and of variance @xmath82 .
thus , there is no need for multiple copies of the filter state and the rest of the algorithm proceeds as before .
this more general algorithm does not perform as well as the algorithm used for local hamiltonians because it searches over a larger hilbert space : the space of the witness and the scratchpad .
this is to be expected since it makes no assumption about the structure of the verification procedure @xmath6 .
note however that all known natural " problems in qma e.g. non - identity check @xcite , consistency of quantum states @xcite , @xmath0-representability @xcite , and zero - error capacity of quantum channels @xcite use a scratchpad of only logarithmic size , so in those cases the running time is the same as for local hamiltonians .
it is tempting to conjecture that the scratchpad of all problems in qma can be reduced to this size . to summarize ,
we have presented a method to prepare ground and thermal states of quantum many - body systems on a quantum computer .
the time required by our algorithm is equal to the square - root of the hilbert space dimension of the system the same time required to prepare the ground state of a classical many - body system .
this represents a speed up by a power of 6 compared to exact diagonalization , which in general is the only available technique to accomplish this task on a classical computer .
it is perhaps surprising that this task can not be accomplished by a straightforward combination of phase estimation and grover s algorithm , but our analysis of this strategy reveals an important failure and more elaborate methods were required .
a quantum computer , if it could be built , could serve as an efficient simulator of quantum many - body systems .
the method we have presented would complement this simulation by initializing the computer in a low - energy state of the simulated system . with some modifications ,
our algorithm can be used to solve and prepare relevant witnesses of all problems in the complexity class qma , the quantum generalization of np . in that case , the physical task consists of preparing a state that has a large overlap with two projectors .
problems in np form a special case where those projectors commute and can be solved straightforwardly using grover s algorithm . however , in the general case the projectors do not commute and more sophisticated techniques were required .
we acknowledge andrew childs , peter hyer , dominik janzing , john preskill , and john watrous for stimulating discussions .
this work began while p.w .
was visiting d.p . who was a postdoctoral scholar at caltech .
24 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , * * , ( ) . , * * , ( ) .
, , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . ,
_ _ ( , , ) . , , , , _ _ ( , ) . , , , * * , ( )
.
, , , , _ _ ( ) , . , , , , , , * * , ( ) . , _ _ , ( ) . , in _ _
( , , ) , .
, , , in _ _ , edited by , , ( , , ) , vol . of _ _ , .
, , , , * * , ( ) . , , , , _ _ ( ) , . , _ _ ( ) , . , , , _
_ , graduate studies in mathematics ( , , ) . , * * , ( ) .
, , , * * , ( ) . , p. ( ) .
, , , * * , ( ) . , _ _ ( ) , . , * * , ( ) . |
the determination of asteroid physical properties is an essential part of the complex process of revealing the nature of the asteroid population . in many cases ,
this process starts with obtaining observational data , continues with creating a model of the asteroid ( i.e. , its size , 3-d shape , and spin state , in the first approximation ) , and ends with interpreting new facts based on the model or a set of these . in this sense
, modeling is a crucial mid - step between observations and theory .
results based on individual well - studied asteroids can be generalized to other members of the population . on the other hand ,
a statistically large sample of asteroids with known properties can reveal physical effects that play an important role for the whole population . in this chapter
, we will build on the content of the _ asteroids iii _
chapter by @xcite about asteroid models reconstructed from _
disk - integrated _ photometry .
although visual photometry still remains the most important data source of the modeling , the main progress in this field since _ asteroids iii _ has been the addition of _ complementary data sources_. many of these data sources are _ disk - resolved _
, thus containing much more information than disk - integrated data .
this shift in paradigm using photometry not alone but simultaneously with complementary data was mentioned in the last paragraph of the _ asteroids iii _ chapter as `` _ perhaps the most interesting future prospect _ '' , and we are now at this stage . in the following , we will review all data types suitable for inversion , their sources , uncertainties , and how they can be used in modeling . when describing the methods of data inversion and the results obtained by these methods , it is also important to emphasize caveats , ambiguities , and possible sources of errors . although the description of what can be obtained from different data sources is exciting , the knowledge of what can not , i.e. , what are limitations of our data sets , is of the same importance .
omitting this may lead to over interpretation of results .
this chapter is structured as follows .
first we review the main principles of the multimodal inverse problem in sect .
[ sec : inversion ] .
then , in sect . [ sec : data ]
, we discuss each data type and their contribution to model characteristics and details , and we describe some extensions of the predominant model . in sect .
[ sec : results ] , we discuss the main results based on lightcurve inversion and multimodal asteroid reconstruction .
we conclude with prospects for the future in sect .
[ sec : future ] .
asteroid physical model reconstruction from multimodal data is , by its nature , a mathematical inverse problem .
it is ill - posed ; i.e. , the uniqueness and stability properties of the solution are usually not very good unless the data are supported by a number of prior constraints .
furthermore , it is not sufficient just to fit some model to the data numerically and try to probe the solution space with some scheme .
although there are more approaches to the problem of asteroid shape reconstruction , they are usually dealing with only one data type and we mention them in the next section . here
, we describe the problem in a general way in the framework of _ generalized projections _ : our data are various 1-d or 2-d projection types of a 3-d model , and understanding the fundamental mathematical properties of the inverse projection mapping is essential .
this includes a number of theorems on uniqueness , information content , and stability properties @xcite .
let the projection point @xmath0 in the image plane ( plane - of - sky or range - doppler ) of the point @xmath1 on the body be mapped by the matrix @xmath2 : @xmath3 . define the set @xmath4 for any @xmath5 as @xmath6 where we have explicitly shown the time @xmath7 and the adjustable parameters : @xmath8 for the shape and @xmath9 for the rotation .
the projection point function @xmath10 if @xmath11 , and @xmath12 otherwise .
the ray - tracing function @xmath13 if @xmath14 is visible ( for occultation , thermal , and radar data ) , or visible and illuminated ( for disk - resolved imaging and photometry in the optical ) ; otherwise @xmath15 .
the set @xmath4 is numerable and finite .
the number of elements in @xmath4 is at most one for plane - of - sky projections ( each point on the projection corresponds to at most one point of the asteroid s surface ) ; for range - doppler , it can be more ( more points on asteroid s surface can have the same distance to the observer and the same relative radial velocity ) .
generalized projections , i.e. , all the data modes presented in sect .
[ sec : data ] , can now be presented as scalar values @xmath16 in the image field @xmath17 : @xmath18 where @xmath19 denotes the luminosity parameters ( for scattering or thermal properties ) , and the luminosity function is denoted by @xmath20 .
the function @xmath21 is the point - spread , pixellation , or other transfer function of the image field . for interferometry , it is typically the fourier transform kernel .
in fact , the reconstruction process works efficiently by taking the fourier transform of any image type rather than using the original pixels @xcite . for lightcurves ,
@xmath22 ( and @xmath5 is irrelevant , @xmath16 is constant ) .
the surface albedo is usually assumed to be constant , although its variegation can be included in @xmath20 by the parameters @xmath19 if there are high - quality disk - resolved data . in the case of lightcurves
only , we can get an indication of non - uniform albedo and compensate for this with a ( non - unique ) spot model @xcite .
the multimodal inverse problem can be expressed as follows .
let us choose as goodness - of - fit measures some functions @xmath23 , @xmath24 , of @xmath25 data modalities .
typically , @xmath26 is the usual @xmath27-fit form between @xmath28 and @xmath29 .
our task is to construct a joint @xmath30 with weighting for each data mode : @xmath31 where @xmath32 denotes the data from the source @xmath33 , @xmath34 is the weight of the source @xmath33 , and @xmath35 is the set of model parameter values .
the best - fit result is obtained by minimizing @xmath30 with nonlinear techniques , typically levenberg - marquardt for efficient convergence .
regularization functions @xmath36 can be added to the sum ; these constrain , for instance , the smoothness of the surface to suppress large variations at small scales , the deviation from principal - axis rotation to force the model to rotate around the shortest inertia axis ( assuming uniform density ) , or the gravitational slope , etc . @xcite
. the modality ( and regularization ) weights @xmath37 are determined using the maximum compatibility estimate ( mce ) principle @xcite .
this yields well - defined unique values that are , in essence , the best compromise between the different data sets that often tend to draw the solution in different directions .
moreover , mce values of weighting parameters are objective , not dependent on users choice , although their values are usually close to those determined subjectively based on experience . plotting various choices of weights
typically results in an l - shaped curve shown in fig .
[ fig_hertha_occ_weight ] ; the best solution is at the corner of the curve . in this way , the reconstruction from complementary data sources is possible even if no single data mode is sufficient for modeling alone . for practical computations ,
the surface is rendered as a polyhedron , and @xmath20 and @xmath38 are computed accordingly with ray - tracing @xcite .
rather than using each vertex as a free parameter , the surface can be represented in a more compact form with spherical harmonics series ( for starlike or octantoid ; i.e. , generalized starlike , shapes ) or subdivision control points @xcite .
these _ shape supports _ are essential for convergence : they allow flexible modifications of the surface with a moderate number of parameters while not getting stuck in local minima or over - emphasizing the role of regularization functions when searching for the best - fit solution .
each shape support has its own characteristic way of representing global and local features .
for example , the octantoid parametrization @xmath39 where @xmath40 , @xmath41 , and @xmath42 are linear combinations of the ( real ) spherical harmonic functions @xmath43 , with coefficients @xmath44 , @xmath45 , and @xmath46 , respectively , is easy to regularize globally while retaining the ability to produce local details .
the coordinates @xmath47 , @xmath48 , @xmath49 , parametrize the surface on the unit sphere @xmath50 but do not represent any physical directions such as polar coordinates .
this inverse problem is a typical example of a case where model and systematic errors dominate over random measurement errors .
thus the stability and error estimation of the solution are best examined by using different model types ( fig . [
fig : daphne_shapes ] ) . in the case of shape , for example
, the reliability of the features on the solution can be checked by comparing the results obtained with two or more shape supports ( starlike , octantoid , subdivision ; * ? ? ?
this yields better estimates than , e.g. , markov chain monte carlo sequences that only investigate random error effects within a single model type .
a particular feature of the model reconstruction from disk - resolved data is that the result is dominated by the target image boundaries rather than the pixel brightness distribution within the target image .
this is because the information is contained in the pixel contrast which is the largest on the boundary ( occultations are special cases of this as they are samples of the boundary contour ) .
this is very advantageous when considering the effect of model errors in luminosity properties ( scattering or thermal models ) : it is sufficient to have a reasonable model , and the result is not sensitive to the parameters @xmath19 .
thus , for example , atacama large millimeter array data can be used for efficient reconstruction even with a very approximate semianalytical fourier - series thermal model
more detailed models have hardly any effect on the shape solution ( sect . [
sec : interferometry ] , * ? ? ?
* ; * ? ? ?
we describe all data types that can be used , the way of collecting the data , their accuracy , typical number of asteroids for which the data exist , and expectations for the future .
we also discuss a typical result of inversion
what is the resolution of the model and how many targets can be modeled ( table [ tab : tech ] ) .
disk - integrated photometry is , and will always be , the most abundant source of data , because it is available for essentially every single known asteroid . because asteroid brightness periodically changes with its rotation , frequency analysis of asteroid lightcurves provides asteroid rotation periods the basic physical property derivable from time - resolved photometry .
the regularly updated asteroid lightcurve database of @xcite available at http://www.minorplanet.info/lightcurvedatabase.html now contains rotation periods and other physical parameters for almost 7000 objects , for about half of them their rotation period is secure and unambiguous .
the role of amateur astronomers in this field is traditionally strong , getting even stronger with increasing level of their technical and software equipment .
hundreds of asteroid lightcurves are published quarterly in the minor planet bulletin ; most of them are then archived in the asteroid light curve database at the minor planet center site ( http://mpc.cfa.harvard.edu//light_curve2/light_curve.php ) in the alcdef standard @xcite .
the efficiency of lightcurve production can be increased by dedicated wide - field photometric surveys ( * ? ? ?
* ; * ? ? ?
* for example ) , although the period determination from undersampled lightcurves is often ambiguous @xcite . for period determination ,
a single lightcurve covering the full rotation is sufficient .
however , a set of such lightcurves observed at different geometries ( asteroid illuminated and seen from various directions ) is needed to reconstruct the shape and spin state of an asteroid . the _ lightcurve inversion _ method of @xcite was already reviewed in _
asteroids iii _ @xcite .
since then , the method has been widely used and hundreds of asteroid models have been derived .
they are publicly available at the database of asteroid models from inversion techniques ( damit , http://astro.troja.mff.cuni.cz/projects/asteroids3d ) .
the reliability of the method was proved by comparing its results with independent data such as laboratory asteroid model @xcite , adaptive - optics images @xcite , stellar occultations @xcite , or spacecraft images of asteroids ( 2867 ) teins @xcite and ( 433 ) eros @xcite . from disk - integrated photometry alone ,
only a global shape without any small - scale details can be derived . because the reflectivity of the surface is not known , the models are not scaled and the information about the size has to come from complementary data . to avoid over - interpretation and artifacts of the modeling ,
the shapes are usually represented by a _
convex _ model .
this allows to work not in the obvious radius parameter space but in the gaussian image space ( describing a convex body by the curvature of its surface ) .
this is less intuitive but it makes the inverse problem less vulnerable to errors of data and model because of the minkowski stability @xcite even if the areas of individual surface facets may change significantly for slightly different data sets , the global convex shape changes very little . from the practical point of view , we are interested in finding a _ unique solution _ of the inverse problem .
to guarantee this , observations covering a sufficiently wide range of viewing and illumination geometries are needed . for a typical main - belt asteroid , it means observations during several apparitions . for a close - approaching near - earth asteroid , several months could be sufficient . for more distant objects , we can in principle
observe changing lightcurve amplitude because of changing aspect , this would anyway take many decades .
nevertheless , trans - neptunian objects ( tnos ) and centaurs can be never observed from earth at geometry significantly different from opposition , which is not sufficient to reconstruct a unique convex model @xcite .
if the lightcurves observed in various filters are different , there is some spectral / color variegation over the surface , and a crude color map can be reconstructed @xcite .
+ + + a typical outcome of lightcurve inversion is a convex shape model that describes the global characteristics of the real asteroid .
it is also easy to obtain nonconvex versions with the general inversion procedures ( see sect . [
sec : inversion ] ) , but then one should produce several solutions with various model types and parameters , and be very cautious about the results @xcite . in general , disk - integrated photometry contains very little information about nonconvexities unless they are very pronounced or observed at very high phase angles where shadowing effects play an important role @xcite .
nonconvex models seldom fit lightcurves better than convex ones simply because the latter typically already fit the data down to noise level .
this sets the resolution limit of photometry ( see the discussion in * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
the relative accuracy of the sidereal rotation period determined from lightcurves is of the order of @xmath51 or better , depending mainly on the time span of observations .
the direction of the spin axis can be determined with an accuracy from a couple of degrees for the models based on many decades of observations with many dense lightcurves , to more than twenty degrees for models based on limited and noisy data . for asteroids orbiting in the ecliptic plane ,
the geometry for an earth - based observer is limited to that plane .
then the disk - integrated brightness of a body with surface described by radius vector @xmath52 and pole direction @xmath53 in ecliptic coordinates is the same as for a body @xmath54 with the spin axis direction @xmath55 @xcite .
that is why for a typical main - belt asteroid , there are usually two equally good mirror shape solutions with about the same pole latitude and pole longitude difference of about 180@xmath56 . this ambiguity can be removed with disk - resolved plane - of - sky projections ( e.g. , images , stellar occultations , see sect .
[ sec : img ] , [ sec : occ ] ) . apart from `` classical '' lightcurves where the sampling of brightness variations is dense with respect to the rotation period
, there are also data that are _ sparse in time_. such data sets typically consist of only a single to a few brightness measurements per night .
provided the whole sparse data set is internally calibrated , it can be used the same way as a standard lightcurve that would be extremely long and very sparsely sampled @xcite .
these data are routinely provided by all - sky astrometric surveys with various usually poor photometric quality .
current surveys provide data of accuracy scarcely better than 0.1mag . given a typical lightcurve amplitude of 0.3mag , the signal is often drowned in noise and systematic errors .
this leads to many physically acceptable models fitting the data to the noise level @xcite . however , the total amount of data is huge and it has been shown that at least for some asteroids , models can be successfully derived from these data or from their combination with dense lightcurves @xcite .
as an example , we show sparse photometry for asteroid ( 21 ) lutetia from us naval observatory ( 260 points covering ten years ) in fig .
[ fig_lutetia_sparse_lc_fit ] and the period search results in fig .
[ fig_lutetia_period_search ] .
the best - fit model based on this data set has the same rotation period and a similar spin axis direction as the models based on much larger and multiple data sets ( fig .
[ fig_lutetia_models ] ) . when using noisy sparse data or only a limited number of data points , a simple shape model of a triaxial ellipsoid is usually sufficient to model the data and to derive the correct period and spin vector orientation @xcite .
the advantage of this approach is that the shape is described by only two parameters ( axes ratios ) and the scanning of the period parameter space is much faster than with general shapes .
the on - going ( pan - starrs , catalina , gaia ) and future ( atlas , lsst ) surveys will provide new data every night for essentially every known asteroid . using this data for automated lightcurve inversion with well - mastered treatment of systematic effects , recognition of the best - fit models , definition of uniqueness of the solution etc . , is the main challenge for the future lightcurve inversion .
the most direct way to obtain information on the shape of an asteroid is to take pictures of it .
the apparent shape as visible on the plane of the sky is delimited by the limb and the terminator , and multiple views obtained while the target rotates can fully characterize its 3-d shape .
to resolve the small angular diameter ( @xmath570.5 ) sustained by asteroids , large facilities are , however , required . in the 1980s , speckle imaging or speckle interferometry provided `` _ _ the first glimpses of an asteroid s surface _
( i.e. , ( 4 ) vesta by * ? ? ?
this technique is based on the analysis of the speckle pattern in the images of astronomical sources obtained through large telescopes at high magnification power and very short integration time .
the aim is to overcome the blurring effects due to the astronomical seeing and to attain diffraction - limited resolution images .
speckle interferometry has been commonly used to study the size , shapes , and surface features of the largest asteroids @xcite . with the launch of the hubble space telescope ( hst ) in orbit and the first light of the large ( 10-m class ) ground - based telescopes equipped with adaptive - optics
fed cameras ( e.g. , w. m. keck , european southern observatory ( eso ) very large telescope ( vlt ) , geminis , subaru ) , the importance of speckle interferometry has decreased .
the critical issue in direct imaging is of course the angular resolution .
any image is the result of the convolution of the object on the plane of the sky with the instrument response , the point spread function ( psf ) . in space ,
the psf is stable and corresponds to the diffraction pattern of the telescope . from the ground
, the atmospheric turbulence constantly deforms the psf and blurs the images , hence the need of real - time correction of the psf by adaptive optics ( ao ) .
the technical challenges of sending a large telescope in space and of building deformable mirrors , explain why the first disk - resolved images in the 1990s were still limited in resolution and only the largest asteroids ( 1 ) ceres , ( 2 ) pallas , and ( 4 ) vesta have been imaged .
in the decade since _
asteroids iii _ , numerous studies have been based on direct imaging of asteroids . both from space or with adaptive optics on the ground , however , the contrast and angular resolution can be improved by deconvolution of the image by the psf .
this is particularly true for images acquired from the ground with residuals from non - perfect ao correction .
deconvolution is an ill - posed problem , but robust algorithms adapted to planetary images are available @xcite and have been validated on sky @xcite .
it is nevertheless the most critical part of the post - processing , as an incorrect deconvolution can introduce a systematic error on the apparent size .
an example of an ao image and the reconstructed shape model for asteroid ( 41 ) daphne is shown in fig .
[ fig : daphne_ao ] current facilities deliver an angular resolution of about 30 to 50 milliarcseconds ( mas ) , depending on the wavelength .
the apparent shape can therefore be measured for asteroids with an apparent diameter larger than @xmath5880100 mas , i.e. , a couple of hundred targets . simulations and observations of known targets such as the satellites of saturn @xcite have shown that a precision of a few mas can be derived on the 2-d profile on the plane of the sky , corresponding to only a few kilometers for main - belt asteroids . with upcoming large telescopes ( 30+m , such as the thirty meter telescope or the european extremely large telescope ) , the angular resolution will be improved by a factor 34 , providing more than 500 targets .
second generation instruments with extreme ao foreseen on these telescopes should allow the observation of about 7000 asteroids , with sizes of only a few kilometers @xcite . using disk - resolved images ,
giant craters have been discovered @xcite , ambiguity in spin solutions have been solved @xcite , albedo maps have been constructed @xcite , convex 3-d shape models from lightcurves have been set to scale @xcite , and full 3-d shape models determined @xcite .
asteroid imaging and mapping _ in situ _ is essentially an extreme example of disk - resolved images .
the modeling in such cases is more a cartographic than an inverse problem as the data are abundant , directly usable ( containing identification points on the surface , etc . ) , and high - resolution ( there is no ill - posedness ) .
however , asteroid _ flybys _ , during which roughly a half of the target is likely to remain unseen ( not visible and illuminated ) , pose a special problem : how to see the dark side ?
the principle here is to use the high - resolution map of the seen side , constructed with a number of methods of photoclinometry , photogrammetry , and image fitting @xcite , as a constraint in the otherwise same multi - mode inverse problem as with ground - based observations .
the procedure is described in @xcite ; with it , the reconstruction of the dark side , such as those of ( 2867 ) teins @xcite or ( 21 ) lutetia ( * ? ? ?
* see also fig .
[ fig_lutetia_models ] ) , is more detailed than from , e.g. , lightcurves alone .
this is because half of the target is accurately reconstructed , with practically no error margin : therefore the fluctuation margin of the dark side is considerably smaller as well .
the observation of a stellar occultation consists in recording the duration of the disappearance of a star behind the asteroid . knowing the apparent motion of the asteroid on the plane of the sky , obtained from its ephemeris , this duration can be converted in a physical length on the disk of the asteroid , called a chord .
provided several observers record the same event from different locations on earth , the 2-d profile of the asteroid is drawn on the plane of the sky @xcite . the main difference with disk - resolved imaging resides in the profile , made by only the limb for occultations , and limb+terminator for imaging . in fine , both techniques provide the 2-d profile of the target as projected on the plane of the sky at the epoch of observations ( fig .
[ fig_hertha_occ ] ) .
disk - resolved imaging and stellar occultation are , however , radically different in term of facilities , data processing , potential targets , reproducibility , and achievable precision . for stellar occultations , the properties of the occulted star matter generally more than the actual target : the asteroid .
if the occulted star is bright enough , its occultation , even by a very small asteroid , can be detected with small aperture telescopes .
this is of course the main advantage of stellar occultations , where the apparent size and shape of potentially any asteroid can be measured .
moreover , this technique can be successfully used also for distant tnos , which have angular sizes too small to be resolved by imaging @xcite . in practice , however , a given asteroid will only seldom occult bright stars .
measurements are thus hard , if not impossible , to reproduce .
stellar occultations are nevertheless extremely valuable .
the accuracy of the timing is dictated by time - series photometry , and can therefore be extremely precise .
an uncertainty of 50ms in timings converts into only a 300 m uncertainty in the length of the chord , typical for a main - belt asteroid ( at 1.5au from earth with an apparent motion of 10/h ) .
the main source of uncertainty is , however , the absolute timing of each chord , required to align them on the plane of the sky .
most of historical occultations were recorded by naked eye , and suffered from this . since a decade ago , thanks to the availability of low - cost positioning and timing systems ( e.g. , gps ) , stellar occultations are being more and more valuable . the main contributors to the field
are currently amateur astronomers : for a given event , observers have to move to set themselves on the predicted occultation path on earth .
small aperture ( @xmath5720 cm ) mobile stations are therefore ideal for recording stellar occultations . because of the uncertainties of the star and asteroid positions on sky
, there is generally an uncertainty of a few tens of kilometers in the location of the occultation path on earth , requiring observers to spread over large area to cover the event .
this usually prohibits dense coverage of the asteroid profile .
when only a couple of chords are available , the event provides only limited information on the size , if any on the shape .
current occultation predictions of sufficient accuracy concern only stars in the hipparcos catalogue and large ( at least tens of kilometers ) asteroids . with the upcoming publication of the esa gaia stellar catalog , and update of asteroid orbits
, this position uncertainty is expected to drop significantly and future occultations will be easier to predict , hence observe @xcite .
overall , stellar occultations can provide precise measurements of the size and shape of an asteroid , as projected on the plane of the sky .
however , events are rare for a given target .
occultations are therefore very valuable in combined data sets , as for instance , to set scale to otherwise dimensionless 3-d shape models ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . from almost 2500 occultations compiled by @xcite , there are about 160 `` good '' ones that allow a reliable determination of asteroid s size and about 40 `` excellent '' ones that show details in asteroid s profile . another technique to overcome the limitations of small angular sizes of asteroids , in order to measure their sizes , shapes , and possible presence of satellites , is interferometry .
an astronomical interferometer combines coherently ( i.e. , conserving the phase information ) the light from two or more apertures of the same telescope or of distinct telescopes spaced by a distance @xmath59 . the spatial resolution ( in radians )
is of the order of @xmath60 where @xmath61 is the wavelength . in the following ,
we give a basic introduction to interferometry of asteroids .
further details can be found in the following works : @xcite and @xcite for a broad introduction to astronomical optical interferometry , methods and instrumentation ; @xcite for a description of the techniques , and the models for deriving the size and basic shape proprieties of asteroids from the eso very large telescope interferometer ( vlti ) mid - infrared interferometric instrument ( midi ) data ; @xcite for an extension of the technique of @xcite to the determination of the sizes and the separation of binary asteroids ; @xcite for a description and the extension of a thermophysical model to the analysis of interferometric data of asteroids with the aim of obtaining surface properties such as the thermal inertia .
+ interferometers measure the coherence function of the source , also called the interferometric visibility , which is given by the ratio between the correlated and the total flux . the correlated flux is the amount of flux in the interferometric fringes .
more precisely the correlated flux is the fourier transform of the brightness distribution of the source measured on the interferometer s baseline projected on the plane of the sky ( see * ? ? ?
* for example , and references therein ) .
the fine guidance sensors ( fgs ) aboard the hubble space telescope ( hst ) are optical white - light shearing interferometers @xcite that combine the light from distinct apertures of the hst primary mirror and have been used to measure the size and basic shape properties of asteroids .
hst / fgs data have clearly demonstrated the bi - lobed nature of some bodies such as ( 216 ) kleopatra @xcite and ( 624 ) hektor , but were not able to detect the presence of the little moons orbiting these asteroids @xcite , due to the large magnitude difference between the asteroids and their respective moons . despite the impressive angular resolution of a few mas , corresponding to a few kilometers at a distance of 1.21.5au ,
a clear limitation of hst / fgs asteroid studies is the bright limiting magnitude of the instrument of about @xmath6214mag .
a recapitulation of the hst / fgs asteroids results can be found in @xcite .
ground - based interferometry is limited to observation of bright targets because of the requirement to take exposures shorter than the atmospheric coherence time of a few milliseconds .
the limiting magnitudes of ground - based long - baseline interferometers and the intrinsic low surface brightness of asteroids , have prevented the use of these instruments for the studies of small solar system bodies until the availability of the vlti .
interferometry in the mid - infrared was proven to be also very sensitive to the global shape of asteroids and also to their surface characteristics @xcite .
in particular , since observations are typically carried out in the thermal infrared ( 813@xmath63 m ) , midi data are sensitive to the surface temperature distribution , which is strongly affected by the value of the thermal inertia ( _ delbo et al .
_ , this volume ) .
figure [ f : interf ] shows the visibility and the total flux as a function of the angular diameter of an asteroid at 2.5au from sun and at 1.5au from earth , where 1mas roughly corresponds to 1 km on the asteroid .
note that main - belt asteroids smaller than @xmath6420 km can not be easily observed with midi at the vlti .
there are about thousand known asteroids with diameter above20 km , implying that interferometry is potentially an interesting technique for shape modelling . on the other hand , as interferometry in the thermal
infrared is sensitive to the spatial distribution of the temperature on the asteroid surface , this technique can be used to determine thermophysical properties of asteroids when the body shapes are known as demonstrated by @xcite .
each vlti baseline can be used with midi only one at the time .
midi will likely be decommissioned in the near future to be substituted after 2016 by the multi aperture mid - infrared spectroscopic experiment ( matisse ) .
this instrument will combine up to four unit telescopes ( uts ) or auxiliary telescopes ( ats ) , allowing six simultaneous baselines .
this feature will enable us to measure the spatial distribution of the infrared flux along different directions .
matisse will also measure closure phase relations and thus offer an efficient capability for image reconstruction .
in addition to the n band , the matisse will also operate in the l and m bands .
unfortunately , matisse is not expected to be more sensitive than midi .
another second generation instrument at the vlti , gravity , that will combine the light from all four uts , will offer further improvements in spatial resolution compared to midi and matisse ( though with more a more stringent @xmath65limiting magnitude ) .
it will provide near - infrared adaptive - optics assisted precision narrow - angle ( about 4 ) astrometry at the 10@xmath63as level in the k band ( 2.2@xmath63 m ) .
both the reference star and the science object have to lie within the @xmath644 field of view . in imaging mode
, gravity can achieve a resolution of @xmath643 mas in the near - ir @xcite .
the imaging mode can be interesting to precisely measure the sizes and the orbits of the satellites of large asteroids , the latter with @xmath65mag .
amber is the current near - infrared focal instrument of the vlti .
it operates in the j , h , and , k bands ( i.e. , from 1.0 to 2.4@xmath63 m ) .
the amber limiting magnitude for asteroid observations is @xmath66mag .
there is only a handful of asteroids brighter than this limit .
these bodies are also the largest ones with angular extensions generally @xmath67100mas , implying a very low visibility in the j - h - k with the uts baselines .
although the photometric flux of these few asteroids is such that @xmath68mag , their correlated magnitudes due to the low visibilities are much higher than 7.5mag , preventing their fringe detection and tracking in the near - ir .
however , one of the most interesting instruments for asteroid ground based interferometry is the large binocular telescope interferometer ( lbti ) .
it consists of two 8.4-m telescopes mounted side by side in a single mount , with a 14.4-m center - to - center spacing .
this configuration offers a unique capability for interferometry of a fizeau beam combination .
this offers a wide field of view ( @xmath641020 ) and low thermal background .
for example , the linc - nirvana instrument can in principle
be able to resolve binary asteroids whose components are separated @xmath69mas .
this will allow splitting many binary asteroids .
another important source of interferometric data in the near future is the atacama large millimeter array ( alma )
. it will provide resolution @xmath645mas at 0.3 mm and a dense mesh of baselines , thus enabling `` imaging '' of hundreds of asteroids in the main belt @xcite . a useful feature of the multimodal inversion is that the raw alma ( and any other interferometric ) data can be used directly as the original fourier transform : there is no need to reconstruct the image estimate .
what is more , the reconstruction of the overall shape is insensitive to inaccuracies and uncertainties in the thermal model used because the model is determined mainly by the boundary of the projection , the distribution of brightness inside the boundary is much less important @xcite .
an example of inversion of simulated alma data is shown in fig .
[ fig_alma ] mas . provided enough observations at different geometries are available ( here eight observation runs were used ) , a shape model ( right ) can be constructed directly from the raw interferometric data.,title="fig : " ] mas . provided enough observations at different geometries
are available ( here eight observation runs were used ) , a shape model ( right ) can be constructed directly from the raw interferometric data.,title="fig : " ] mas .
provided enough observations at different geometries are available ( here eight observation runs were used ) , a shape model ( right ) can be constructed directly from the raw interferometric data.,title="fig : " ] mas .
provided enough observations at different geometries are available ( here eight observation runs were used ) , a shape model ( right ) can be constructed directly from the raw interferometric data.,title="fig : " ] long - baseline interferometry can also represent a novel approach to determine the masses and the densities of asteroids in a range of sizes and distances never studied before .
in particular , modern interferometers such as the vlti , lbti and the magdalena ridge observatory interferometer can spatially resolve binary asteroids discovered by photometric lightcurves in the main belt @xcite .
the separation of the components of these systems is too narrow for traditional observational techniques such as ao at 10-m class telescopes . the esa gaia mission , whose operations started in 2014 , will provide accurate astrometry and photometry of asteroids .
photometric data can be used the same way as sparse photometry from ground - based observatories ( sect .
[ sec : photometry ] ) .
the potential of gaia - like data was demonstrated by @xcite on data from the hipparcos satellite .
apart from disk - integrated brightness , gaia will provide also direct measurements of asteroid sizes , in a way much similar to the interferometry .
being designed as an astrometry mission at the @xmath63as level , the psf of gaia is accurately known and stable .
deviations of the psf from that of a point - like source can thus be measured and used to estimate the apparent size of asteroids .
owing to the amount of sources observed by gaia , the satellite does not download images to earth , but only 1-d flux profiles , corresponding to small 2-d windows centered on targets , stacked along one direction , similar to interferometry in that respect .
interpretation of this 1-d profile relies therefore on the _ a priori _ knowledge of the 2-d flux distribution on the plane of the sky . in the creation of gaia catalog ,
this will be done iteratively , starting from simple spheres at zero phase angle to finally use the spin and tri - axial ellipsoid solutions determined otherwise from gaia photometry @xcite .
these measurements represent a great opportunity for multi - data inversion algorithms : the measurement of the extension is direct , but clearly requires a realistic description of the projected shape on the plane of the sky .
considering gaia specifications and observing geometry , @xcite have estimated the fraction of asteroids for which 1-d dimension will be measured with a precision better than 10% .
this fraction is highly dependent on diameter , and 20% of asteroids between 2030 km will be measured at least once , while all asteroids larger than @xmath6480 km will be measured repeatedly .
measurements of asteroids emission in the thermal infrared in general at wavelengths between 45 and 2030@xmath63 m are mostly used to determine the sizes of these bodies .
ground based telescopes can only observe in specific windows of the electromagnetic spectrum where the atmosphere is relatively transparent : i.e. , the l , m , n , and q bands at 3 , 5 , 10 , and 20@xmath63 m , respectively . however , such observations are strongly affected by the variability of the transparency of the atmosphere and its thermal background .
the background also receives contribution from the telescope and the optics .
hence , the thermal infrared observation of asteroids from the ground is limited to relatively bright asteroids ( @xmath70mag with 10-m class telescopes ) .
absolute calibration of the flux is rarely better than 510% . on the other hand , in space
the instrument calibration is usually stable and there is no need to reduce the thermal background from the atmosphere . as a consequence , space based telescopes such as
spitzer can observe much fainter and smaller asteroids @xcite , with uncertainties in the calibration that can reach @xmath641% error .
also , from space the range of the observational wavelength is limited only by the detector technology , typically @xmath643.550@xmath63 m . at longer wavelengths ,
the telescope optics require cooling and the observation of faint objects is confronted with the background from solar system dust cloud and infrared cirrus . in the _ asteroids iii _ era
, the main source of thermal infrared observations of asteroids has been the iras minor planet survey that collected observations of more than 2200 asteroids @xcite . since then
, the nasa wide - field infrared survey explorer ( wise ) has observed more than 130,000 main - belt asteroids ( _ mainzer et al .
_ , this volume ; * ? ?
* ; * ? ? ?
* ) , about 500 near - earth asteroids @xcite , about 1100 hilda asteroids @xcite , and almost 2000 jupiter - trojan asteroids @xcite in four infrared wavelengths at 3.4 , 4.6 , 12 , and 22@xmath63 m ; the akari space telescope observed more than 5000 asteroids during its mission @xcite ; the spitzer space telescope observed hundreds of asteroids ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* among others ) ; the herschel space observatory , that , due to its longer wavelengths , spanning 55671@xmath63 m , was primarily used to observe trans - neptunian objects @xcite . for a review about all these missions and their results see _ mainzer et al .
_ , this volume .
the thermal infrared spectrum of asteroids carries information about their size and surface properties , such as the thermal inertia , roughness and emissivity .
these properties are typically derived by interpreting thermal infrared data by means of thermal models .
the `` simple '' thermal models that assume a spherical shape , a lambertian emission of the surface , and a simplified calculation of the surface temperature distribution are used when we lack knowledge of the asteroid global shape , spin vector and rotation period , which is the majority of the cases .
widely used are , for example , the near - earth asteroid thermal model , or the standard thermal model .
however , in order to derive the thermal inertia of an asteroid from measurements of its thermal infrared emission , more sophisticated models , called thermophysical models ( tpms ) , are needed .
such models are used to calculate the temperature distribution over the body s surface as a function of different parameters , including the thermal inertia . in these models ,
the asteroid shape is usually fixed and is modeled as a mesh of planar facets .
the temperature of each facet is determined by numerically solving the one - dimensional heat diffusion equation using assumed values of the thermal inertia , with the boundary condition given by the time - dependent solar energy absorbed at the surface of the facet ( see _ delbo et al .
_ , this volume , for a review ) .
this latter quantity is calculated from the heliocentric distance of the asteroid , the value assumed for the albedo , and the solar incident angle .
macroscopic surface roughness is usually modelled by adding hemispherical section craters of variable opening angle and variable surface density to each facet . shadowing and multiple reflections of incident solar and thermally emitted radiation inside craters are taken into account as described by @xcite and .
heat conduction is also accounted for within craters @xcite .
surface roughness can be adjusted by changing the opening angle of the craters , the density of the crater distribution , or a combination of both @xcite .
the total observable thermal emission is calculated by summing the contributions from each facet visible to the observer .
model parameters are adjusted until the best agreement with observational data is obtained , i.e. , the least - squares residual of the fit is minimized , thereby constraining the physical properties ( albedo , size , macroscopic roughness , and thermal inertia ) of the asteroid . from the point of view of multi - data inversion however , the optimization of thermophysical parameters as described above is a two - step process
first , the spin and shape model is derived from one data type ( photometry , radar , ) , then this model is fixed and used for deriving thermophysical parameters from another data type ( thermal infrared ) .
this approach lacks the possibility to weight the two data types with respect to each other .
moreover , the thermophysical parameters can be very sensitive to small modifications of the input shape and spin , so various modifications of the shape should be tested to see how stable the solution is @xcite . ideally , one should model shape and spin parameters together with thermal parameters .
this multi - data approach using lightcurves and thermal infrared data simultaneously was successfully tested by @xcite and in principle it can be used also to data that are sparse in time .
radar observations that measure the distribution of echo power in time delay and doppler frequency ( so - called range - doppler or delay - doppler measurements ) are discussed in detail in @xcite and in _ benner et al . _
( this volume ) .
the delay - doppler projection is many - to - one mapping of a 3-d surface of the target into a 2-d `` image '' .
each pixel on the image represents a bin containing integrated echo power from surface elements that have the same distance from the radar and the same relative speed ( due to the rotation of the asteroid ) . from the point of view of inversion ,
images in the range - doppler plane are generalized projections that can automatically be handled with the general procedure discussed in sect .
[ sec : inversion ] and in detail in @xcite and @xcite . in this approach ,
the multi - mode reconstruction is tuned to produce models with intermediate scale resolution ( @xmath641/10 of the diameter ) since these are computationally inexpensive ( can be obtained in a few minutes with a laptop ) , and data sources other than radar do not contain more detailed information .
an example of radar range - doppler data of asteroid 2000 et@xmath71 is shown in fig .
[ fig : radar_data ] and the corresponding reconstructed shape model in fig .
[ fig : radar_shape ] . if detailed radar data are available , such model
can then be further refined @xcite with the radar techniques described by _
benner et al . _
( this volume ) . due to the steep decrease of echo power with the distance to the object ( the fourth power of the distance ) , asteroids achievable by current radar facilities arecibo and goldstone are only close - approaching near - earth asteroids or the largest members of the main belt .
a number of examples of what can be reconstructed from radar data is given by _
benner et al . _
( this volume ) .
we briefly discuss here the interesting possibility of peer into an asteroid .
this is somewhat separate from the remote - sensing framework as the data can only be obtained in situ ; on the other hand , such data can be acquired with the future space missions .
the most practical option is to plant radio transmitters / receivers on the surface of a kilometer - sized asteroid and measure low - frequency ( @xmath64100mhz ) signals between these and an orbiter when they pass through the interior of the target .
the most robust observables are simply signal travel - time data @xcite .
these allow an efficient formulation of the inverse problem via the refraction index and are relatively insensitive to noise and model error .
nevertheless , they suffice to give a coarse - scale picture of the general distribution of permittivity inside the asteroid , as well as the locations and sizes of large anomalies ( sudden low- or high - density regions such as voids or heavier minerals ) .
this approach has also been robustly tested in laboratory conditions @xcite .
a more refined possibility is to measure changes in the pulse profile , although this is more prone to errors @xcite .
the interior of the asteroid is practically impossible to model accurately in three dimension since it is supposed to have a number of cracks , voids , discontinuities etc .
, all refracting and reflecting the radio waves in complicated ways .
thus a very robust scheme is essential for extracting the available information with stability .
regardless of the data type , one or two transmitters on the surface are not sufficient for a unique solution .
a tetrahedral configuration of four transmitters would be ideal , but this places heavy demands on the payload design .
so far , we assumed that the asteroid can be described as a solid single body with constant spin vector , i.e. , rotating along the axis with the maximum moment of inertia with a constant rotation rate .
although this model represents a typical asteroid , there are other configurations that can be also treated with an extension of the simple model .
binary asteroids form a significant part of the population of small asteroids .
@xcite estimated that the fraction of binaries is 15% for the near - earth population and a similar fraction is assumed for the main - belt population in the same size range ( _ margot et al .
_ , this volume ) .
the formation , dynamics , and physical properties of binary and multiple asteroid systems are discussed in detail in other chapters of this volume ( _ walsh et al .
, margot et al . _ ) . in general , modeling of such systems is more complex because the dynamics has to be taken into account .
however , in some cases described below , the technique used for modeling single asteroids can be used also for multiple systems or at least their primary components . for binary ( or multiple ) systems where the primary component is much larger than the satellite ,
the photometric signal from the satellite can be neglected and the primary can be modeled as a single body . for such systems , the shape of the primary and the constraints on its gravitational quadrupole @xmath72 from the orbit analysis of the secondary can be used to investigate the distribution of the density @xcite . for systems with comparable sizes ,
the problem becomes complicated when the system is asynchronous , i.e. , when the rotation period of the primary is different from the orbital period of the secondary .
such systems are usually modeled as two ellipsoids @xcite , although more general models were created from radar observation ( 1999 kw@xmath73 , for example , see * ? ? ?
fully synchronous binaries can be approximated by single bodies if the separation of their components is not large .
even a convex model can provide a good fit to the lightcurves @xcite .
such model does not represent the true configuration of the system , of course , but it provides the correct rotation period and direction of the orbital plane .
when the separation of components is larger , the system has to be modeled as two - component .
however , from the modeling point of view , it is just a moderate modification of the nonconvex problem , where the system is described by only one rotation / orbital period and orientation of the normal of the orbital plane ( parallel to the spins of the bodies ) .
if the model is based on lightcurves only , the spin and period parameters can be reconstructed accurately , but the uncertainty in shapes is large . as has been shown by @xcite on ( 624 ) hektor , the distinction between a highly nonconvex single body , two bodies in contact , or two bodies orbiting each other is difficult to make .
an example of reconstruction of a doubly synchronous binary system ( 90 ) antiope from lightcurves and occultations is shown in fig .
[ fig_antiope ] , where the model is shown together with the silhouettes from occultations .
tens of chords observed during the occultation in 2011 @xcite portray the two components to details unattainable by any other observational technique and the large set of lightcurves observed over many apparitions constrains the rotational parameters .
a similar model can be obtained also by using lightcurves separately to create a scale - free model that is then scaled by occultation data ( the shaping asteroids with genetic evolution algorithm , * ? ? ?
. however , this two - step approach lacks the advantages of simultaneous inversion where the two data types can be weighted with respect to each other . as described in detail by _ vokrouhlick et al . _
( this volume ) , rotation state of small asteroids is affected by the anisotropic recoil of scattered sunlight and thermal radiation , which causes a net torque called the yarkovsky - okeefe - radzievskii - paddack ( yorp ) effect .
this effect secularly changes the obliquity of the spin vector and the rotation period .
whereas the former is too small to be measured with current data , the latter has been measured on several asteroids ( see table 3 in _ vokrouhlick et al . _ , this volume ) .
if the change of the rotation period is larger than the uncertainty of the period , the change can be traced from apparition to apparition as was the case for ( 54509 ) yorp @xcite . in other cases ,
the effect was much smaller and it revealed itself by the discrepancy between the data and the model assuming the period to be constant @xcite .
the yorp effect is easy to include into the model .
we assume that the rotation rate @xmath74 changes linearly in time @xmath7 as @xmath75 .
then the parameter @xmath76 is another free parameter of the modeling .
because the shift in the rotation phase increases quadratically in time as @xmath77 , even small changes @xmath78 of the order @xmath79rad / d@xmath80 can be detected with data sets covering tens of years . in principle , the measured value of @xmath76 can be compared with the theoretical value computed from the spin state , shape , size , and the thermal parameters of the surface with the density as a free parameter . however , due to sensitivity of yorp on small scale details of the shape that are far below the resolution of the model @xcite , and the problem of transverse heat diffusion @xcite , this can hardly be more than a rough comparison . because the yorp effect scales as inverse of the square of the size of asteroid , it becomes more important for small bodies , where it might be necessary to include it into the modeling if the data cover a wider span of time .
yorp is assumed to play an important role in many dynamical processes the distribution of rotation periods and spin obliquities of small asteroids @xcite or the creation of asteroid binaries and asteroid pairs @xcite , for example .
it is important to have more asteroids with yorp detection .
better statistics of values of period change will help to constrain theories of yorp evolution of small asteroids .
asteroids rotating in the relaxed mode are fully described by the spin axis direction , rotation rate , and the initial orientation
. however , some asteroids are in an excited rotation state , which can be described as a rotating free top .
more parameters are needed to describe this tumbling motion @xcite .
the reason why some asteroids are in this state can be ( i ) primordial , ( ii ) collisional excitation @xcite , or ( iii ) end state of yorp - driven spin down .
the approach to the modeling is in principle the same as for asteroids in principal axis rotation , only the orientation for a given time is given by solving differential equations . from lightcurves , models of asteroids 2008 tc@xmath81 @xcite and ( 99942 ) apophis @xcite have been derived . a model of asteroid ( 4179 ) toutatis @xcite
was derived from radar data and lightcurves .
this model was later compared with the fly - by images of change-2 mission
the general shape was in agreement , although there were some minor discrepancies @xcite .
software for the inverse problems is available at damit ( http://astro.troja.mff.cuni.cz/projects/asteroids3d ) .
convexinv is a procedure for lightcurve inversion , while adam ( all - data asteroid modeling ) is a collection of functions from which one can tailor an inversion procedure for any data sources @xcite . an earlier version of this is called koala ( knitted occultation , adaptive optics and lightcurve analysis , * ? ? ?
* ) ; this is based on lightcurves and silhouette contours obtainable from images and occultations @xcite .
koala is especially suitable for lightcurve and occultation data , while adam allows the use of any images ( camera , radar , or interferometry ) with or without lightcurves without having to process them to extract contours or other information .
both koala and adam can be used for lightcurves only but , as discussed earlier , this is not reliable .
any resulting shape should only be taken in the global sense ( as a more realistic - looking rendering of a convex solution ) , and the details or nonconvex features are seldom likely to be real @xcite .
the purpose of modeling methods described in previous sections is to reveal new facts about the nature of asteroids .
the approach is made on two fronts : the first strategy is to use the most abundant data sources ( photometry in visual and thermal ir ) to produce many low - resolution models that will be a statistically significant sample of the whole asteroid population from which conclusions about the physical properties can be drawn .
the second approach is to concentrate on selected targets , obtain as many different data types as possible , create detailed models of these asteroids , and extrapolate the obtained results to the whole class of similar objects .
we describe in this section new research areas that directly benefit from availability of spin solutions and 3-d shape models .
one of the main results of the lightcurve inversion is the increasing sample of asteroids with known orientation of the spin axis . for main belt asteroids ,
the long - known lack of asteroids with poles close to the ecliptic @xcite was confirmed and it was shown that it is more pronounced for smaller asteroids @xcite . in fig .
[ fig_beta_vs_size ] , we plot the distribution of pole latitudes for @xmath64800 asteroid models with respect to their size ( an updated version of fig . 5 in * ?
the size - dependent structure can be explained by the yorp effect that is more effective on smaller asteroids ( @xmath8230 km ) and pushes them into extreme values of obliquity ( 0@xmath56 or 180@xmath56 ) .
this corresponds to the clustering of pole latitudes towards values of @xmath8390@xmath56 . although there are observation and modeling biases that affect the distribution of poles in the sample of available models , their effect is only marginal compared to the strong anisotropy seen in fig .
[ fig_beta_vs_size ] @xcite
. the spin - axis orientation of even smaller asteroids ( @xmath825 km ) is still not known due to the lack of models .
for the largest asteroids ( @xmath8460 km ) , there is a statistically significant increase of prograde rotators ( 98 prograde vs. 63 retrograde in fig .
[ fig_beta_vs_size ] ) , probably of primordial origin @xcite .
a different approach to the problem of spin - axis distribution was used by @xcite .
they analyzed variations of the mean brightness with the ecliptic longitude , from which they estimated ecliptic longitudes of spin axis for about 350,000 asteroids and revealed a clearly non - uniform distribution .
however , the explanation of the cause for this non - uniformity is still missing . with increasing sample of models , it is also possible to study the distribution of spin axes of members of collisional families .
results of @xcite agree with theoretical expectations : if the spread in proper semimajor axis increases with decreasing size due to the yarkovsky effect ( _ nesvorn at al .
_ , _ vokrouhlick et al .
_ , this volume ) , asteroids closer to sun than the center of the family should rotate retrograde , those farther should rotate prograde . this behavior is shown in fig .
[ fig_flora ] for flora family . unlike most astronomical objects , the viewing geometry of asteroids continuously changes due to their motion relative to observer . as a corollary , knowing their rotation period and spin - vector coordinates is crucial to interpret projected size measurements , and to tie together observations . as already described in sect .
[ sec : data ] , the spin and 3-d shape are required to interpret thermal infrared radiometry , stellar occultations , or apparent sizes measured by gaia without biases . similarly , there have been long running discussions to explain
inconsistent spectral measurements of asteroids , that can be easily solved once the spin properties ( period and orientation ) are known : see , for instance , the discussion on ( 832 ) karin by @xcite , @xcite , and @xcite , or on ( 21 ) lutetia by @xcite .
+ density is one of the most fundamental properties to constrain the composition of asteroids and investigate their internal structure ( _ scheeres et al .
_ , this volume ) . with the exception of binary asteroids with observable
mutual eclipses ( _ margot et al . _ , this volume ) , both mass and volume are required to determine the density of an asteroid . estimating the mass of an asteroid by measuring its gravitational influence on other objects is a challenge because of the relative low mass of asteroids compared to other planetary objects .
the number of mass determinations thus limits the number of density estimates .
although there are diameter , hence volume , estimates for all asteroids with a mass determination , the uncertainty in volume generally dominates the balance on density uncertainty @xcite . in this respect ,
3-d shape models are required to determine accurate volume .
the level of potential biases increases with stronger assumptions on the shape , and accuracy accounting for systematics improves from sphere , to ellipsoid , to convex hull , to the real shape .
in the decade since _
asteroids iii _ , the number of density estimates has increased from 20 to 300 objects @xcite . among these ,
the most reliable are derived from binary systems for which the volume of the primary was determined after shape modeling .
a detailed description on the density of asteroids and their internal structure can be found in _
scheeres et al . _
( this volume ) , from under - dense asteroids , hosting large voids , to over - dense asteroids , likely differentiated . in the context of this chapter
, it is important to highlight than only density estimates more precise than 1020% can be used to discriminate between different analogue meteorites and can provide insights on the internal structure . such accuracy can only be achieved with volume known to 510% or better , which means that a proper description of the 3-d shape is needed .
the spin state and shape model of an asteroid is input information for the tpms . while in the _ asteroids iii _ era ,
shape and spin information were available only for a handful of asteroids , preventing application of tpms to a large number of these bodies , this situation has drastically changed in the last few years .
physical properties ( such as the value of the surface thermal inertia ) of about 60 asteroids are now available thanks to the application of tpms ( see _ delbo et al .
_ , this volume ) . a remarkable improvement in this field is also represented by the availability of high quality thermal infrared data as those produced but the wise and the spitzer space telescopes . in the next few years we expect the number of asteroids with known thermal inertia values to grow thanks to the availability of more shape and spin state models
. from the spin , 3-d shape , and density ,
the local gravity at the surface can be computed .
unsuspected physics has been unveiled with the modeling of the near - earth asteroid ( 66391 ) 1999 kw@xmath73 by @xcite .
some small asteroids present an equatorial bulge , presumably generated by regolith migration toward lower gravity regions .
this process can even form binary systems if the asteroids spin fast enough @xcite .
+ an evident outcome of shape modeling is the capability to detect large impact craters and basins . aside from the spacecraft encounters ,
the first detection was the large impact basin on vesta , progenitor of the vestoids @xcite , detected with the hst @xcite and confirmed by nasa dawn spacecraft @xcite .
another case is the recent impact suffered by asteroids ( 596 ) scheila , detected by the presence of a dust tail @xcite .
lightcurves obtained before and after the impact , under similar geometries are different , revealing different surface properties @xcite .
+ lcccccrrr photometry & x & x & & x & & 30 & 500 & @xmath85 + images & & x & x & x & & 5 & 50 & @xmath86 + occultation & & x & x & x & & 5@xmath87 & 50 & @xmath86 + radar & x & x & x & x & & 10 & 30 & @xmath86 + radiometry & & & x & & x & 10@xmath87 & 20 & @xmath85 + interferometry & & x & x & x & & 5@xmath88 & @xmath89 & @xmath86 + fly - by & x & x & x & x & x & 6 & 10 & @xmath90 + because some asteroids are less or more dense than their most - likely constituents , the question of the mass distribution ( denser material or voids ) in their interior can be asked .
this question is intrinsically tied with the study of the gravity field around the asteroid .
the latter has been measured during spacecraft encounters ( see * ? ? ? * for instance ) , but studies from earth - bound observations have recently appeared .
by comparing the spherical harmonics of the gravity field as determined from the orbit of a natural satellite , with the expected coefficients resulting from the 3-d shape model , the hypothesis of homogeneous mass distribution can be tested @xcite .
in the decade since _ asteroids iii _ , where the principles of lightcurve inversion based on dense - in - time series were presented @xcite , the number of models has seen a tenfold increase from a few tens to a few hundreds ( table [ tab : tech ] ) . the increasing availability of sparse - in - time photometry , and its appropriate handling in the inversion , coupled with a dramatic increase of computer time ( thanks to projects like asteroids@home ) have made this possible .
upcoming all - sky surveys such as pan - starrs , lsst , and gaia are expected to produce enormous data sets and there is little doubt that thousands of models will be derived in the next decade .
our knowledge on non - gravitational effects such as yorp and yarkovsky will directly benefit from this larger sample . with the large number of data and models
, new challenges will arise how to extract scientifically interesting information from a large set of models of asteroids ?
with big data flows , the processing has to be automated , with effective data processing .
the obvious search for correlations has to be done with care because of large biases in the set of models .
although the importance of detailed models of individual well - studied asteroids will be important , the main shift in paradigm and probably the main source of interesting findings will be in tens of thousands of asteroid models derived from photometry in optical and thermal infrared wavelengths . only a few years ago
, only photometry in the visible was available for a large number of asteroids .
the situation drastically changed with wise catalog of thermal fluxes for 150,000 asteroids .
automatized procedures capable of dealing with photometry in the visible and thermal infrared will yield not only 3-d shape and spin state for thousands of asteroids , but also their diameter , albedo , and thermal inertia of their surface .
the later being crucial in interpreting asteroid mineralogy once coupled with spectroscopy ( _ reddy et al .
_ , this volume ) .
understanding of observational and modeling biases will be crucial for correct interpretation of the results .
connecting spin and shape distribution of asteroids with their orbital and spin evolution will hopefully lead to a clear picture of the evolution of the main asteroid belt .
the ultimate goal here is the connection of models of evolution of the main belt with spin - axis evolution and current distribution of asteroid physical properties .
another approach that is complementary to modeling individual objects , is modeling distribution functions of parameters of interest . with any inversion technique and photometric data quality
, the number of models will be always much lower than the number of known asteroids just because it takes time to collect enough data at different geometries .
so instead of aiming to create unique models for a statistically significant sample of the ( sub)populations with known observational and modeling bias , one can use other observables than the time brightness pairs . with this approach
, there is almost no `` wasting '' of data because essentially all photometric points are used in the statistical approach .
the aim is to model characteristics of a given asteroid population as a whole when there are not enough data to model individual members .
one of the observables can be , for example , the mean brightness over one apparition @xcite and its dispersion that correlates with lightcurve amplitude .
if there are not enough data points to estimate mean brightness and its variance , the statistics of scatter of individual pairs of observations can be used @xcite .
if the data come in pairs of observations separated by a constant time interval ( like with pan - starrs ) , the rotation period can be estimated @xcite .
another challenge for next decade resides in the derivation of a large sample of higher - resolution models , based on multiple data sources .
building inversion techniques capable of handling different data sources had been highlighted in _
asteroids iii _ as the next step , and this became reality over the past few years .
the models based on multiple data sources are more realistic , more reliable , more precise than the independent analysis of individual data sets .
if practical issues of merging data sets arise , mathematical solution have been exposed , and algorithms and software are ready and freely available .
the current sample of models derived by such methods is nevertheless still limited .
if the multi data analysis is more efficient than the independent analysis of the individual data sets , the practical problem of data sharing arises .
although the tools to invert multiple data sources are ready and freely available , the data are still hard to get and harder to share . to maximize the scientific output ,
it is necessary to combine all available data sources .
the other observing techniques ( i.e. , disk - resolved imaging , stellar occultations , gaia size measurements , or radar echoes ) will never provide data for more than a few thousands asteroids .
the apparent sizes by gaia will however be measured with a high precision , and will concern a decent sized sample .
because the data will be made public , they should be used on a regular basis for modeling .
the number and quality of profiles derived from stellar occultations have always been restricted due to the intrinsic complexity of prediction , and cost of equipment .
the parallel availability of low - cost telescopes and cameras , together with the publication of gaia stellar catalog and asteroid orbits , will open a new age for stellar occultations .
any mass - production procedure of asteroid models should therefore be able to deal not only with photometry ( visible and infrared ) but also with stellar occultations . finally ,
if the bulk of asteroids are single objects , rotating along their shortest axis , a small fraction will either be in tumbling state , or have satellites .
automatic modeling procedures should be able to detect such cases , triggering detailed analysis of these peculiar targets .
the rising interest of private companies for near - earth objects as mining resources will inevitably lead to further development of analysis of remote - sensing data , building dedicated ground - based or space telescopes , and eventually moving from asteroids surface into their interiors . *
acknowledgments . *
the work of j was supported by the grant gacr p209/10/0537 of the czech science foundation .
md acknowledges support from the grant anr-11-bs56 - 008 shocks of the french national research agency .
j. , scheirich p. , kaasalainen m. , et al .
( 2007 ) . .
in _ near earth objects , our celestial neighbors : opportunity and risk _ ( milani a. , valsecchi g. b. , and vokrouhlick d. , editors ) , p. 191 . cambridge university press , cambridge .
e. p. , lupie o. l. , mcarthur b. , et al .
( 1998 ) . .
in _ astronomical interferometry _ ( reasenberg r. d. , editor ) , volume 3350 of _ society of photo - optical instrumentation engineers ( spie ) conference series _ , pp . | = 11pt |
young massive ob stars significantly influence the structure , dynamics , chemistry and thermal balance of their associated molecular clouds through the impact of their ultraviolet photons .
their extreme ultraviolet photons ( euv , @xmath11 13.6 ev ) ionize the gas immediately surrounding them , forming regions . far ultraviolet photons ( fuv , 6 ev @xmath12 13.6 ev ) dissociate molecular gas beyond the region , creating a photodissociation region or pdr ( hollenbach & tielens 1999 ) .
pdrs are ubiquitous : fuv photons dominate the gas heating in pdrs and affect the chemistry and physics of gas over a large fraction of the volume and mass of molecular clouds .
the study of interactions between fuv radiation and molecular cloud gas is therefore important in understanding molecular cloud evolution and the feedback between massive star formation and subsequent star formation in molecular clouds .
observations of pdrs compared with theoretical models probe the physical conditions in star - forming regions through dust and gas emission mainly in the infrared ( ir ) wavelength region of the spectrum .
pdrs are the source of most of the ir emission in the galaxy .
dust grains and large carbonaceous molecules such as polycyclic aromatic hydrocarbons ( pahs ) absorb radiation from the stars , and re - radiate this energy flux in the infrared , producing a continuum with solid state and pah spectral signatures .
fuv photons incident on dust grains and pahs also cause photoelectric ejection of electrons which then collisionally heat the gas , a process called the grain photoelectric heating mechanism ( e.g. , watson 1972 , bakes & tielens 1994 , weingartner & draine 2001 ) .
the gas cools via emission in many ir lines .
emission from pdrs has been quite successfully modelled ( see hollenbach & tielens 1999 and references therein ) .
the models are usually parameterized by @xmath13 , the ratio of the fuv flux to the habing ( 1968 ) fuv band flux of @xmath14 erg @xmath3s@xmath15 characteristic of the local interstellar radiation field , and @xmath16 , the density of the gas . for advecting ( non - stationary ) models ,
the flow velocity of the material through the pdr is also a parameter .
the fuv fields in pdrs near ob star - forming regions such as the orion bar or m17 sw are typically very high , ( @xmath13 @xmath17 @xmath18 ) ; the pdr gas in these regions is inferred to be very dense with average densities @xmath19 @xmath20 ( tielens & hollenbach 1985b ) . on larger scales and in more evolved regions such as the rosette nebula , the fuv fields and the average densities may be significantly lower by factors of @xmath21 .
infrared , sub - millimeter and millimeter wavelength observations of pdrs indicate that gas in pdrs is inhomogeneous ( van der werf et al .
1993 , luhman et al .
1998 ) , which is not surprising since molecular clouds themselves are observed to be highly inhomogeneous on a wide range of scales from tens of parsecs down to tenths of a parsec ( e.g. , genzel 1991 ) .
there is both indirect and direct evidence for clumpiness in pdrs .
spatially extended fine structure line emission of neutral and singly - ionized carbon near regions shows that the fuv penetrates deeper into the cloud than predicted by homogeneous models , suggesting that the gas is clumpy in nature ( meixner et al .
1992 , stutzki et al .
1988 , steiman - cameron et al .
further , strengths of high excitation lines of molecular species like co j=14@xmath2213 are observed to be greater than that predicted to arise from gas with density inferred from other pdr species .
it is thus inferred that in regions such as m17 sw and the orion bar , the pdr gas consists of a `` low '' density ( typically @xmath23@xmath20 ) component responsible for the extended ci and cii emission and high - density clumps ( n @xmath24 @xmath20 ) which give rise to the high excitation co lines from their surfaces or , more precisely , at a depth of @xmath251 from their surfaces ( burton et al .
1990 , meixner et al .
1992 , steiman - cameron et al .
1997 , hogerheijde et al .
clumps are also observed directly by mapping pdrs at high spatial resolution in the 2 @xmath26 lines of h@xmath27 ( e.g. , orion bar , luhman et al .
1998 ; eagle nebula , allen et al .
the small regions of enhanced emission are interpreted as indicating the presence of high density clumps ( @xmath28 @xmath20 ) with sizes @xmath29 0.02 pc , occupying a very small volume filling factor @xmath30 . however , this interpretation is not very conclusive , and in this paper we present evidence to show that small clumps with such high densities may not exist in pdrs .
small differences in temperature within the observed regions can also explain the observed emission , as h@xmath27 emission is highly sensitive to gas temperature ( marconi et al .
1998 ) .
dense pdrs like the orion bar are spatially thin structures and if these pdrs are indeed clumpy , the clumps must have sizes smaller than the thickness of the warm pdr , @xmath31 pc for the orion bar .
the presence of small clumps affects the infrared spectrum from pdrs by introducing a range of densities in the fuv - illuminated region and by introducing enhanced advection in clumps that are photoevaporating .
one of the goals of this paper is to investigate the conditions under which small , high - density clumps exist in pdrs .
small clumps might not exist in significant numbers if the intense fuv field photoevaporates them quickly .
the incident fuv field heats up and pressurizes the surface layers of the clumps , causing the layers to expand .
this mass loss may cause a complete photoevaporation of the dense clumps , and thus destroy them on short timescales .
the mass loss timescales due to photoevaporation of an fuv - exposed clump can be estimated by assuming that the heated gas at the surface flows outwards at its sound speed , @xmath32 .
the mass density at the base of this heated flow , @xmath33 , is lower than the mass density , @xmath34 , in the cold clump , as we will discuss below .
( see table 1 for a list of frequently used symbols in this paper . )
the mass loss rate is given by @xmath35 and since @xmath36 , the photoevaporation timescale is @xmath37 . for a typical clump , and an fuv - heated surface at 1000k ,
this is approximately @xmath38 years .
asymmetric mass loss from the surfaces of very small clumps can rocket them ( e.g. , oort & spitzer 1955 ) out of the pdr back into the cloud .
clumps that survive the rocket effect and remain in the pdr are likely to lose significant fractions of their mass on short timescales of order @xmath39 years , depending on the ratio @xmath40 .
the clump evaporation timescale can be longer or shorter than its residence time in the fuv - illuminated surface region of a gmc . from a frame of reference moving with the advancing ionization front , interclump material and the clumps in the gmc
are advected into the pdr and ultimately flow into the region .
large clumps survive and they can penetrate the region and affect its evolution and structure , becoming evaporating gaseous globules or `` eggs '' ( bertoldi & mckee 1990 ) . in the present paper , we focus on the heating of an individual clump in a pdr by fuv photons from a nearby o or b star .
we study a range of fuv fluxes incident on a range of clump sizes and determine the photoevaporative lifetimes of the clumps and the evolution of their structure .
our investigation is analogous to the study by bertoldi ( 1989 ) and bertoldi and mckee ( 1990 ) on the effects of euv heating of clumps in regions , and a generalization of the studies by johnstone et al .
( 1998 ) and strzer & hollenbach ( 1999 ) of the evaporation of small clumps and protoplanetary disks ( `` proplyds '' , odell et al . 1993 ) by euv and fuv photons .
in a subsequent paper ( gorti & hollenbach , in preparation ) we will discuss the cumulative effect of o and b stars on giant molecular clouds ( gmcs ) , and the relative importance of euv and fuv photons in dissociating and destroying gmcs .
ultimately , we are interested in the role of fuv radiation in regulating star - formation in gmcs .
destruction through photodissociation and photoevaporation of potentially star - forming clumps / cores in gmcs reduces the star - forming efficiency of gmcs , and the fuv field may thereby play an important role in regulating or limiting star formation .
photoevaporation of the outer envelope of a collapsing isothermal sphere by fuv radiation could rapidly deplete the material in the envelope , and may limit the final mass of the star formed . on the other hand , compression of non - collapsing clumps by shock waves driven by the warm surface gas could possibly drive the inner cores to instability and gravitational collapse , triggering star formation . in 2 of this paper ,
we discuss the present understanding of gmc structure and existing models . in 3 , we describe the evolution of an region and its surrounding pdr as they propagate into the gmc .
timescales relevant for the fuv heating of a clump in various contexts are briefly discussed in 4 . in
5 we present a simple analytical model which provides an understanding of the underlying physics of fuv - induced clump evolution .
we consider various possible evolutionary scenarios and their dependences on clump parameters and the strength of the fuv field . for the more general case ,
a 1-d numerical hydrodynamics code has been developed and these results are described ( 6 ) . in 7 , we estimate the significance of the rocket effect due to asymmetric mass loss and its effect on clump lifetimes in the pdr , and in 8 we discuss the implications of our results for observations of pdrs in star - forming regions .
application to observed pdrs is made in 9 .
we conclude with a summary of the present investigation ( 10 ) .list of symbols frequently used in this paper [ cols="^ , < " , ]
giant molecular clouds are observed to have a hierarchical substructure that is highly inhomogeneous , with density enhancements such as cores , clumps and filaments on all observable scales , from @xmath41 pc down to @xmath42 pc .
we use the term `` massive cores '' to denote large scale density enhancements ( @xmath43 pc in size ) and `` clumps '' to denote smaller scale structures ( @xmath17 few tenths of pc or less in size ) .
gmcs and their more massive constituent cores are strongly self - gravitating , and are supported against collapse by turbulence and magnetic fields .
the internal velocity dispersions of the substructures in molecular clouds are found to scale with their sizes , with decreasing non - thermal support on the smallest scales ( goodman et al .
the column density through a gmc or a typical gmc substructure is found to be approximately a constant @xmath44 @xmath3 , although significant variations ( @xmath45 @xmath3 ) about this typical column may occur .
( for a recent review on gmcs , see mckee 1999 ) .
the smaller substructures or clumps are not all gravitationally bound and the exact physical nature of these clumps remains uncertain ( williams et al .
they have been interpreted as temporary fluctuations in density caused by supersonic turbulence within the cloud ( e.g. , falgarone & phillips 1990 , scalo 1990 ) .
this view is supported by their apparent fractal nature , which seems to show the same structure on smaller scales down to a few tenths of a parsec ( elmegreen 1999 ) , and their observed supersonic linewidths ( blitz & stark 1986 ) . in this picture ,
the gravitationally unbound clumps are transient objects formed by converging supersonic turbulent flows .
the non - gravitating clumps are therefore being constantly formed and destroyed , and have short lifetimes of the order of their sound crossing time , @xmath46 , where @xmath47 is the clump sound speed , and @xmath48 is the initial clump radius .
the effect of the sudden turn - on of an fuv field on such clumps is to drive a shock into the clump , producing a smaller , denser clump with a warm evaporative outflow .
surprisingly , the evaporative timescale is now somewhat longer than @xmath49 , as we will show . assuming that the fuv flux does not strongly affect the turbulence so that the formation timescale remains the same , the presence of an fuv field will not greatly change the non - gravitationally bound clump abundance , but it will compress them and alter their thermal structure and dynamical evolution .
another interpretation is that small bound clumps as well as small unbound clumps represent stable physical entities confined by interclump pressure .
observations through molecular line spatial maps and velocity channel maps or position - velocity diagrams indicate the presence of dense coherent structures .
indirect evidence suggests the presence of a low density ( @xmath50 orders of magnitude less than the clumps ) `` interclump medium '' ( icm ) ( see williams et al . 1995 and references therein ) .
the icm exhibits high velocity motions ( @xmath51 km s@xmath15 ) and therefore may have significant pressure , which could be of thermal , magnetic or turbulent origin .
clumps in this picture are confined by the icm , and some of them may be self - gravitating ( maloney 1988 , bertoldi & mckee 1992 ) .
pressure - confined clumps live much longer and presumably form more slowly than the gravitationally unbound clumps of the turbulent model . therefore
, fuv heating of these clumps and their destruction through photoevaporation will have a much more pronounced effect on lowering the clump abundance in this interpretation .
massive stars form in the central regions of clumpy , massive , cores ( @xmath52 ) , of radius @xmath53 pc inside gmcs ( see review by evans 1999 ) .
these massive cores contain clumps whose densities may reach @xmath54 @xmath20 , or even higher ( plume et al 1997 ) , and the clumps are surrounded by interclump gas with lower , but uncertain densities , possibly @xmath55 @xmath20 .
when a massive star forms , the euv photons from the star immediately ionize the nearby interclump gas , and form an region .
the clumps in the region are also exposed to the ionizing radiation and evaporate ( bertoldi 1989 , bertoldi and mckee 1990 ) .
the fuv photons penetrate further through the interclump gas beyond the interclump ionization front ( if ) and dissociate the molecular gas , forming a dissociation front ( df ) and a pdr . in the earliest stages of evolution when euv photons emerge from the forming o / b star , the if moves as an r - type front
. however , the if quickly stalls and becomes d - type , at which point the high pressure region drives a @xmath56 10 km s@xmath15 shock into the neutral interclump gas beyond the if .
it is instructive to obtain a sense of sizescale at the r / d type transition .
assuming the massive star emits @xmath57 lyman continuum photons per second , the strmgren radius at this juncture is @xmath58pc , where @xmath59s@xmath15 and the electron number density @xmath60 @xmath20 . if the central star has an fuv luminosity of @xmath61 ( typical of stars with @xmath62 ) , the fuv flux incident on the surrounding pdr is characterized by @xmath63 .
the thickness of the pdr is approximately @xmath64@xmath65 or @xmath66 pc .
the df initially advances ahead of the shock , but as photodissociation approaches equilibrium with h@xmath27 formation , the df slows down . eventually , the shock moves ahead of the df , and the entire pdr is contained in the post - shock shell .
the shock now impacts cold molecular interclump gas .
hill & hollenbach ( 1978 ) showed that at this stage the shock velocity has slowed to @xmath67 km s@xmath15 , and that , if @xmath68 @xmath20 , the shock completely stalls in @xmath69 years .
therefore , there is an initial stage ( @xmath70 years , for dense icm ) in the evolution of the clumpy pdrs surrounding regions where the clumps are both shocked and exposed to the photoevaporating effects of the fuv radiation .
this can be followed by a stage where the shock has stalled and unshocked clumps move into the fuv - illuminated zone .
two configurations are possible after the shock stalls . if the region is still completely embedded , it becomes stationary with its pressure matched by the molecular cloud pressure and the ionizing photons completely absorbed by recombining electron / proton pairs in the region .
much more likely is the situation where the region breaks through the surface of the cloud and becomes a `` blister '' region . in this case , the ionized gas can expand away from the cloud , and into the interstellar medium .
the if then slowly eats its way into the pdr and the df similarly advances into the molecular cloud .
the velocity of the if advancing into the pdr is of order @xmath71 km s@xmath15 , just enough so that the flux of particles through the if can balance the heavily attenuated flux of euv photons reaching the if ( see whitworth 1979 , bertoldi & draine 1996 , strzer & hollenbach 1998 ) . in such a case , the clumps entering the pdr may not have been previously shocked and compressed .
further , in the turbulent gmc model , clumps form and dissipate continually and those formed in the pdr after the passage of the shock are not affected by it . in both the pdr around the trapezium stars in orion and in the m17 sw pdr ( meixner et al .
1992 ) , there is no evidence for velocity shifts between pdr gas and the ambient cloud gas , or , in other words , no evidence for a shock at the present time . in this paper
, we focus only on the photoevaporation of clumps , whether or not they have been shocked , and do not follow the potential shredding or flattening of a clump by the passage of a shock wave ( see , e.g. klein et al .
1994 ) .
the thickness or column depth of a pdr has been defined by tielens & hollenbach ( 1985 ) to include all the gas and dust where fuv photons play a significant role in the gas heating or chemistry . by this definition , in regions like orion or m17 , the pdr extends to columns of @xmath72 @xmath3 and includes gas which is cool and molecular ( h@xmath27 , co ) , but where fuv still photodissociates o@xmath27 and h@xmath27o .
we are interested in the surface column n@xmath73 of the pdr which is heated to the highest temperatures ( t@xmath74k , depending on @xmath13 and @xmath16 ) . for g@xmath75 @xmath76 , n@xmath77 @xmath3 ( equivalent to a dust fuv optical depth of order unity ) and is nearly equal to the column where atomic hydrogen converts to h@xmath27 in stationary pdrs ( tielens & hollenbach 1985 ) .
we shall call this region the `` warm pdr '' hereafter to distinguish it from the entire pdr .
the thickness of the warm pdr , n@xmath78 , is of order @xmath79 pc .
the crossing timescale for the if through the warm pdr is @xmath80 years , where @xmath81 km s@xmath15 ) .
gas and dust ( and clumps ) from the opaque molecular cloud interior are therefore advected on these timescales from the shielded cloud , through the warm pdr , and across the if into the region .
figure 1 shows a schematic diagram of fuv radiation from nearby o / b stars impinging on massive cores ( large sizescales , e.g. , rosette ) and clumps ( small scales , e.g. , orion ) .
fuv - illuminated clumps are heated at their surfaces on timescales @xmath6 that could be either fast or slow compared to the internal sound crossing timescale within the clump , @xmath82 .
impulsive heating implies that the fuv heating of the clump surface layer , characterized by column @xmath83 , occurs with @xmath84 . in this case
, the outer region is heated more quickly than the entire clump can respond , and attains a much higher pressure than the cold central region .
a shock propagates inward and a heated photoevaporative flow expands off the surface .
several situations in clumpy molecular clouds may lead to @xmath85 . for a clump with @xmath86 cm and @xmath87 km s@xmath15 ( t @xmath17 10k ) , @xmath88 years .
the ob star may `` turn - on '' on a timescale @xmath89 years .
even after this initial flash of fuv flux , the clumps may shadow each other so that @xmath6 is given by the time for a clump to move out of shadow , which is of order @xmath90 , where @xmath91 is the clump velocity . in a turbulent
, supersonic medium @xmath91 may be larger than @xmath47 so that @xmath92 .
similarly , if clumps are formed in turbulent gas by supersonic converging flows , the fuv turn - on timescale is roughly the time for the clump to form , which again is @xmath93 , where @xmath94 is the supersonic turbulent speed .
therefore , impulsive heating is likely the most appropriate approximation for the turbulent model of clumps in molecular gas .
let us consider the other situation of long - lived clumps confined by the icm . here , slow heating is generally the best approximation , but in certain conditions , impulsive heating may be more appropriate
. a common situation after the turn - on of the ob star may be the advancement of the ionization front and pdr front into the opaque molecular cloud , which occurs at speeds @xmath95 km s@xmath15 .
in this situation @xmath6 is the time for the df and the if to move so that the clump is one fuv optical depth closer to the if , or @xmath96 , where @xmath97 is the thickness of the heated pdr zone in the icm of density @xmath98 .
the criterion for impulsive heating @xmath99 , is more easily met by larger clumps that have longer sound crossing timescales , as @xmath6 is the same for all clumps .
this gives us a limiting size for clumps likely to be impulsively heated , @xmath100 which is of the order of the thickness of the interclump pdr region itself . clumps with radii smaller than this are slowly heated as they emerge from the opaque molecular cloud into the fuv - irradiated regions . since we only consider clumps _ within _ pdrs that have @xmath101 , pressure - confined clumps are generally slowly heated in our analysis .
clumps that undergo slow heating , @xmath102 , adjust quasi - statically to the changing fuv flux .
long - lived , small , pressure - confined clumps advecting into fuv - irradiated regions provide a prime example of this slow evolution .
such clumps develop a warm pdr surface which slowly gets warmer , thicker and less dense . the clump and the heated pdr surface evolve in near pressure equilibrium , with the density of the surface decreasing with the rise in its temperature , such that @xmath103 pressure of the icm ( assuming thermal pressure dominates on these small scales ) . at any given time
, the evolutionary state of a slowly heated clump is the same as the _ final _ state of an impulsively heated clump exposed to the same local fuv flux .
however , a gradually heated clump constantly adjusts its pdr surface to maintain pressure equilibrium as the pdr slowly heats .
in contrast , an impulsively heated clump undergoes shocks and supersonic expanding flows in its attempt to attain equilibrium .
( we note that there is no final equilibrium state if the interclump pressure is zero ) .
rapid photoevaporative mass loss can cause an acceleration due to a rocket effect in very small clumps and move them back into the shielded cloud .
pressure - confined small clumps , on the other hand , are not rocketed as they evolve quasi - statically . in a given fuv field and at a given icm pressure ,
the final equilibrium state of a small clump is either a completely photodissociated , warm @xmath104 , and expanded ( but with @xmath105 ) region ; or it may consist of a shielded cold molecular core with the same temperature , density and pressure as the original clump , but with a heated and more diffuse protective surface layer of pdr material . in
5.3 we provide analytic solutions for these equilibrium structures
. however , in 5 and 7 we primarily focus on the interesting physical processes which occur for clumps that are impulsively heated .
we first provide approximate analytic solutions for the time evolution of a clump suddenly exposed to fuv radiation , using a few simplifying assumptions . as described in 2 ,
the nature of the clumps themselves is not well understood , and we consider two simple analytic models that qualitatively correspond to the two main interpretations .
turbulent unbound clumps are modelled as constant density structures in vacuum .
a clump in vacuum disperses on a sound crossing timescale , in the absence of fuv illumination .
pressure - confined clumps are also treated as being of constant density , but have a surrounding icm with equal pressure . in our models ,
the icm only serves to confine the clumps and does not itself evolve or otherwise affect the clump evolution .
we seek solutions for the time evolution of the mass , size and mass loss rate of a photoevaporating clump .
clumps are assumed to be dense , small spheres of gas , supported by thermal , turbulent and magnetic pressures .
we assume that the magnetic field @xmath106 scales with a constant power of the density , so that the magnetic pressure is @xmath107 the initial ratio of turbulent and magnetic pressures , @xmath108 and @xmath109 , are set by two dimensionless parameters , @xmath110 where @xmath111 is the initial thermal gas pressure in the clump
. a clump may have significant turbulent support and observations of clumps in star - forming regions ( jijina et al .
1999 ) indicate values of @xmath112 ranging from @xmath113 in cold , dark clouds to @xmath114 in regions of massive star formation .
there is observational evidence to suggest that turbulent support in clumps decreases on the smaller scales of star - forming clumps ( @xmath115 pc ) , where @xmath116 ( goodman et al .
. measurements of magnetic fields and hence @xmath117 and @xmath118 are difficult , but present observational data suggest a wide range of values for @xmath117 , from @xmath113 to a few ( crutcher 1999 ) .
the value of @xmath118 ( @xmath50 ) describes how the b field responds to a relatively sudden change in gas density . for a frozen , uniform magnetic field threading a spherical clump , @xmath119 or @xmath120 . for our standard case
we adopt values of @xmath121 , and @xmath122 .
we later discuss the sensitivity of our results to these choices for @xmath112 , @xmath117 and @xmath118 .
as discussed in 4 , for clumps confined by an icm and advecting into fuv - illuminated interclump zones , the assumption of @xmath123 is equivalent to @xmath124 .
therefore , in this case we treat only clumps above a certain minimum size , which is of order the size of the thickness of the heated pdr of the icm .
we do not explicitly include self - gravity in our analytical model , and only qualitatively estimate the effects of gravity on clump evolution .
we treat gravity more quantitatively in the numerical models .
the fuv radiation field as seen by the clump is spherically asymmetric , and stronger at the surface facing the source . for simplicity
, however , we assume a spherically symmetric fuv field when calculating the internal dynamical evolution and the thermal structure of the evaporating clump . typical albedos of interstellar dust are @xmath125 and some of this scattered radiation is directed backwards ( dependent on the phase factor @xmath126 , see henyey & greenstein 1941 ) , resulting in a backscattered fuv flux on the shielded clump surface about @xmath127 times that on the directly illuminated surface ( e.g. , dsert et al .
1990 , hurwitz et al .
although the posterior intensity is diminished compared to that on the front surface , the pdr surface sound speed is not a sensitive function of the incident fuv flux @xmath13 ( kaufman et al .
1999 ) . for field strengths @xmath128 near typical regions and densities @xmath129 @xmath20 characteristic of clumps there , even a back flux 10% of that on the clump surface facing the source would heat the gas to nearly the same sound speed ( @xmath130 km s@xmath15 on the source side
, @xmath131 km s@xmath15 on the shadowed side ) .
therefore , the assumption of spherical symmetry is as good as ignoring the flux of fuv photons on the backside when calculating the thermal structure and internal dynamics .
however , the asymmetry of the flows from the front and back surface will cause a rocket effect on the clumps and induce motion away from the source , as first recognized by oort & spitzer ( 1955 ) .
we discuss momentum transfer due to mass loss of the clumps in 7 , and estimate sizes of clumps that gain velocities high enough to move out of the pdr and back into the molecular cloud . in our analytical models ( but not our numerical models ) we assume that the transition from the warm outer pdr layer to the cold molecular core of a clump is very thin , so that the clump can be modelled as an isothermal core with sound speed @xmath47 , surrounded by a pdr envelope of sound speed @xmath32 .
the gas in the core always remains isothermal in our analysis , even when shock - compressed to high densities . for our pressure - confined clumps
, it is assumed that the icm is of low density and does not absorb any fuv photons . in order to obtain analytic solutions ,
we do not fully treat transient phenomena and emphasise quasi - steady state aspects of the flow .
we begin by considering the evolution of a simple configuration : a cloud consisting of clumps that are spheres of radius @xmath132 and constant initial density @xmath133 , immersed in a vacuum .
we refer to such clumps hereafter as `` turbulent clumps '' .
these are dynamic structures that in the absence of fuv heating expand at roughly their sound speed , @xmath47 .
when such a clump is exposed to an fuv field , the photon flux is attenuated by dust and instantaneously heats an outer column of n@xmath134 @xmath3 ( thickness @xmath135 ) to relatively high temperatures and sound speeds ( for example , near regions with @xmath136 and @xmath137 @xmath20 , @xmath138k and @xmath139 km s@xmath15 ) .
the evolution of the clump is determined by its initial radial column density from the centre outwards , @xmath140 , and the ratio @xmath4 of sound speeds in the fuv - heated region and the cold clump material , @xmath141 the parameter @xmath4 can be thought of as a measure of the strength of the fuv field incident upon the cold clump .
we introduce the dimensionless parameter @xmath1 , @xmath142 for the initial column density to the centre of the clump .
note that @xmath1 is a measure of the mean column density through a clump . for turbulent clumps with @xmath143 ,
the initial clump column density is less than @xmath83 , and the entire clump is immediately heated and photodissociated by the fuv flux .
effectively , therefore , the fuv flux accelerates the expansion of the cold clump by heating it throughout , decreasing the expansion timescale by a factor given by the ratio of sound speeds at the two temperatures , @xmath144 , or @xmath145 . for clumps with @xmath146
, the fuv flux heats an outer shell of gas to a higher temperature , thereby increasing its pressure .
if the pressure of the outer pdr shell is sufficiently high , it drives a shock to the centre of the clump rapidly compressing it , and the compressed clump proceeds to evaporate on a somewhat longer timescale than the expansion timescale in the absence of an fuv field . on the other hand ,
if the pressure in the heated outer pdr shell is low , the shock does not make it to the centre of the clump .
the shock dissipates , followed by an expansion of the clump , until finally the photoevaporative flow halts the expansion and proceeds to shrink and compress the gas .
there are thus two distinct evolutionary scenarios for photoevaporating turbulent clumps with @xmath146 . for brevity in the main text ,
the details of our analysis and the derivations of the results are presented in the appendices to the paper .
we now discuss our solutions and describe the physics of fuv - heated clump evolution and refer the interested reader to the appendices for the full analysis .
clumps exposed to a given fuv field , measured by the parameter @xmath4 , are either compressed by shocks or expand to adjust their column density to a critical value , @xmath147 .
this critical column density , @xmath147 can be derived from conservation of mass and from the condition of pressure equilibrium at the surface of the cold clump gas ( see appendix a for details ) , and is defined by the relation @xmath148 recall that @xmath112 and @xmath117 are the initial ratios of turbulent to thermal pressure and magnetic to thermal pressure , respectively , and that the magnetic pressure scales as @xmath149 . for typical values of these parameters ,
the second term on the lhs is negligible and equation ( [ aetext ] ) can be used to define a photoevaporation parameter , @xmath150 , as @xmath151 for clumps with initial column densities @xmath152 , the photoevaporation parameter @xmath153 .
figure 2 shows an @xmath154 parameter plot with equation ( [ aetext ] ) depicted for the standard case with @xmath121 , and @xmath120 , and for two possible extremes of these parameters .
the result is not very sensitive to the choice of the parameter @xmath118 , and @xmath147 only depends on the sum of @xmath112 and @xmath117 . for the standard case and with @xmath155 and @xmath156 which is typical of many pdrs , equation ( [ aetext ] ) can be approximated as @xmath157 .
the evolution of fuv - heated clumps is determined by where they initially lie in the @xmath154 parameter space . clumps with an initial column density @xmath158 evolve through a shock - compression phase , until their dimensionless column density increases to @xmath147 ( see appendix b ) .
clumps with higher initial column densities expand until their column density decreases to @xmath147 ( see appendix c ) .
the evolution of clumps in two representative cases is qualitatively shown in figure 2 through their trajectories in @xmath154 space . once the clump column density attains its critical value @xmath147 , further evolution is similar in both cases , and the clump column density at any later time remains constant at this critical value .
we note that this would imply that observed clumps in a given pdr should typically have the same column density , the critical value @xmath147 given by the ambient fuv field .
this is valid for clumps photoevaporating into a vacuum and only violated in the transient early stages which are typically very short , of order @xmath159 . from equation ( [ aetext ] ) , clumps at 10k heated on their surfaces to 100 k ( or @xmath160 ) should have columns @xmath161 @xmath3 , and for @xmath162 or @xmath163 @xmath3 .
as photoevaporation proceeds , and clumps lose mass , they shrink in size and their average density increases , so that the column through the clump remains the same , @xmath164 .
[ [ shock - compression ] ] shock compression + + + + + + + + + + + + + + + + + in this case , the initial column density is larger than @xmath83 , but smaller than @xmath165 .
the fuv can only penetrate through an outer shell of thickness @xmath166 .
this shell is heated to a higher temperature and the corresponding increase in pressure causes the outer shell to expand at @xmath32 radially outwards at the edge and inwards into the clump at the inner boundary .
figure 3 shows a schematic diagram of the evolution of a shock - compressed clump .
as the shell expands , the column density @xmath83 through it is maintained by further penetration of fuv photons , and heating of more cold clump material .
the pressure at the base of the pdr shell is initially higher than the cold clump gas .
this pressure difference drives a shock wave into the clump , which rapidly propagates to the centre . behind the shock
the cold clump gas is compressed to a pressure approximately equal to the pressure at the base of the pdr shell .
the shock travels inward at roughly @xmath32 and the entire clump gets compressed in a time @xmath167 , where @xmath168 is the initial radius of the cold clump when the fuv flux is turned on ( fig .
the shock reaches the centre and compresses the clump to a radius @xmath169 .
the shock - compressed gas forms a high - density core , such that the pressures in the core and at the base of the pdr flow are approximately equal .
further mass loss from the clump is from the surface of this core .
the time evolution of the mass and radius of the clump for @xmath170 are given by the following equations for our standard case , where @xmath171 , @xmath172 , and @xmath120 .
for the general solution , refer to appendix b. @xmath173 @xmath174 @xmath175 here @xmath176 , @xmath177 and @xmath178 . the pdr
/ core interface propagates inward as the warm gas evaporates and , finally , the entire core is transformed to heated , photodissociated , expanding pdr material . using equations ( [ r2simtext ] ) and ( [ m2simtext ] ) ,
figure 4 shows the clump radius and mass as a function of time for a case where the initial column through the clump is given by @xmath179 , and @xmath180 represents the incident fuv flux .
this is a typical column density for clumps in molecular clouds , ( see 2 ) and for a clump of mass @xmath181 , corresponds to a density @xmath182 @xmath20 . in this example of @xmath180 ,
the fuv field might heat cold clump gas with initial temperature of @xmath5630 k to about @xmath56750 k , or @xmath183k gas to about 250 k. also shown in figure 4 are the results of a more detailed hydrodynamics code simulation ( described in the next section ) for the same parameters .
there is very good agreement between the two results , in spite of the various simplifying assumptions made in arriving at the analytical solutions .
the numerical solution for @xmath184 oscillates with time due to the dynamics of the shock compression which we have neglected in the above analysis .
our analytic solution assumes a quasi - steady state , whereas the numerical code shows the effect of overshooting the steady state solution and rebounding .
the clump evolution goes through two distinct phases , which is apparent from the plot of clump radius with time ( fig.[4a ] ) .
the clump initially shrinks rapidly with time , as the shock compresses the clump in a time @xmath185 .
the average velocity @xmath186 with which the radius decreases is @xmath187 for the chosen parameters .
after the entire clump is compressed by the shock , it decreases in size more slowly , now due entirely to the mass loss from its surface . as discussed earlier , the column density in the compressed clump remains constant , and the clump gets denser as its size decreases . in figure 4b , the mass of the clump is also seen to decrease with time as the entire clump gets photoevaporated in about a sound crossing time ; again , the analytical results closely follow the result of the numerical hydrodynamical code .
the cold clump mass at @xmath188 is less than the initial clump mass because it excludes the mass contained in the outer column @xmath83 which is instantaneously heated by the fuv flux to the pdr temperature . using the results of the above analysis
, we can obtain simple estimates of photoevaporation timescales for a clump exposed to incident fuv radiation .
we define the photoevaporation timescale , @xmath189 , as the time for the radius of the clump to shrink to zero .
the time @xmath189 , can be easily determined from setting the lhs of equation ( [ r2simtext ] ) to zero , @xmath190 or @xmath191 larger clumps are longer - lived compared to smaller clumps in a given pdr environment .
photoevaporation timescales of turbulent clumps in pdrs in typical star - forming regions are thus of the order of a few clump sound crossing times ( see 7 ) .
paradoxically , the sudden turn - on of fuv radiation on a clump in a vacuum can increase its lifetime , even though the fuv heats the surface and increases the flow speed from the surface ! for turbulent clumps with lifetimes @xmath192 in the absence of an fuv field ,
exposure to the fuv field results in a `` pressure - confined '' period where the clump is compressed due to pressure of the heated surface pdr layer .
this compression reduces the area of the photoevaporating surface and can extend the lifetimes of these clumps to a few @xmath49 .
[ [ collapse - of - clumps - driven - by - shock - compression ] ] collapse of clumps driven by shock compression + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the fuv driven shock wave that compresses a clump to very high densities , may render it gravitationally unstable to collapse .
thus star formation may be triggered in a previously stable clump .
a simple estimate of parameters leading to clump collapse can be made without explicitly including self - gravity in the equations .
we use our solutions for the radius of the shock - compressed core ( appendix b ) and compare this with its jeans length , @xmath193 . if @xmath194 , the clump is stable , otherwise it undergoes gravitational collapse .
it should be noted here that a clump supported by magnetic pressure with @xmath195 will always be stable to collapse regardless of the external pressure , as the magnetic pressure increases faster during compression than the gravitational energy density .
our discussion here is thus applicable for clumps with insignificant magnetic fields , or with a magnetic equation of state where @xmath196 .
we now solve for the collapse criterion as a function of initial clump parameters and the fuv field strength as measured by @xmath4 , for an illustrative case where @xmath121 , and @xmath197 .
the radius of the shock - compressed core is given by @xmath198 the jeans length of the compressed core is given by @xmath199 for collapse @xmath200 which yields @xmath201 the factor @xmath202 in equation ( [ collapse ] ) is the initial thermal pressure in the clump , and for clumps in pdrs is typically of the order of @xmath203 @xmath20 k. for @xmath204 @xmath20 k , equation ( [ collapse ] ) becomes @xmath205 figure 5 shows the collapse criterion(eq.[[collapse1e6 ] ] ) on the @xmath154 parameter plot .
clumps with initial column densities greater than @xmath206 and with initial thermal pressures of @xmath207 @xmath20 k can not support themselves against gravity and are unstable to collapse even in the absence of an fuv field .
this region of instability is to the right of the vertical dotted line in the plot .
clumps to the left of this line are initially stable , but sufficiently strong fuv fields may drive shocks that compress them and trigger collapse .
the fuv fields near ob stars may thus trigger star formation in previously stable clumps , and _ increase _ the star formation rate in clouds , in cases where @xmath208 .
parameter space and demarcates the regions where gravitational collapse can be triggered by shock compression ( for @xmath209 ) .
the dashed line indicates the column densities of clumps , with thermal pressure @xmath207 @xmath20 k , which are initially unstable to collapse .
clumps with this pressure and columns to the left of the line are initially stable .
the regions to the right of the solid line are where clumps are driven to collapse by shocks induced by photoevaporation , for initial thermal pressures in the clump @xmath210 . for higher pressures @xmath211 ,
both the dotted ( @xmath212 ) and solid lines ( @xmath213 ) move to the right . ]
[ [ clumps - with - an - initial - expansion - phase ] ] clumps with an initial expansion phase + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + clumps with initial column densities @xmath214 as given by equation ( [ aetext ] ) develop very thin ( @xmath215 ) pdr shells when they are first exposed to fuv radiation .
this very thin shell is initially at a high pressure and drives a shock into the clump on the inside edge of the shell while the outside edge of the shell expands at roughly @xmath32 just as is the case for shock - compressed clumps .
however , the shock stalls before it reaches the clump centre as the pdr pressure rapidly declines due to the shell expansion .
the pdr pressure and clump pressure become equal before the shell thickness is comparable to the clump radius .
we summarize below the detailed evolution of initially expanding clumps ; details appear in appendix c. because of the high pdr sound speed and relatively small thickness , the pdr pressure initially drops faster than the pressure in the cold clump .
the pressure in the cold clump eventually becomes higher than that in the surrounding expanding pdr shell , and the cold clump gas expands at its sound speed , @xmath47 ( as it would in the absence of fuv radiation since the interclump medium is assumed here to be a vacuum ) .
this expansion takes place until a time @xmath216 at which the density ( and pressure ) in the clump gas drops to that of the outer pdr layer . for @xmath217 ,
fuv photons begin penetrating into the formerly shielded cold clump gas .
the warm pdr gas now confines the clump gas , and further expansion of the cold clump is halted .
the cold clump proceeds to lose mass gradually and shrinks due to photoevaporation , with an evolution similar to the final solution for shock - compressed clumps . the radius and mass of the clump at times @xmath217 and for @xmath218 are given by @xmath219 where @xmath220 , @xmath221 and @xmath222 figure 6 shows the change in radius and mass of a clump with initial parameters @xmath223 .
the clump expands out initially and loses a small fraction of its mass during this phase .
after its density has dropped , so that there is pressure equilibrium between the cold clump gas and the heated layer immediately surrounding it , expansion is halted . as photoevaporation causes mass loss off its surface , the clump shrinks in size , and is completely heated and photodissociated in about 4 crossing timescales .
the numerical results are also overlaid on the plot , and there is reasonable agreement .
the photoevaporation timescale for initially expanding clumps can be defined analogously to shock - compressed clumps as ( see appendix c , eq.([tpeexp ] ) ) @xmath224 therefore in this regime of low @xmath4 , the photoevaporation timescale @xmath189 is of order several sound crossing timescales , and is proportional to @xmath225 . for a cloud model in which clumps are structures in pressure equilibrium with an icm ,
the evolution of fuv - heated clumps is somewhat different .
the heated shell or pdr can now no longer expand indefinitely , but is eventually confined by the pressure of the icm .
the evolution of the clump is again determined by @xmath1 and @xmath4 .
clumps with an initial column density @xmath226 , are completely heated and photodissociated instantly .
such clumps expand until the pdr gas reaches the interclump pressure .
clumps with higher initial column densities again are only heated on their surfaces to a column @xmath83 , and their further evolution depends on the turn - on time of the fuv field @xmath6 , relative to @xmath49 .
if the clumps are heated impulsively , @xmath227 , pressure - confined clumps initially evolve similarly to the turbulent clumps discussed earlier .
clumps with @xmath158 are thus shock - compressed , and mass flows are set up which photoevaporate the clump .
however , the outflow runs up against the interclump pressure and eventually reaches pressure equilibrium with its surroundings . during this time
, the clump may either be completely heated through and get transformed into a sphere of warm pdr gas , or may only be partially photoevaporated .
the final configuration for partially photoevaporated clumps consists of a small remnant cold clump , and an extended warm pdr envelope surrounding it and protecting it from the fuv flux .
the clump , pdr envelope and interclump medium are in pressure equilibrium .
clumps with @xmath214 are not much affected by the fuv field .
there is no expansion of the clump due to the presence of the confining icm , and the initially heated thin pdr shell expands to form an expanded , but still thin , protective pdr layer .
containment of the pdr shell by the icm prevents further penetration of fuv photons into the cold clump and there is no additional heating . for these clumps
, photoevaporation is relatively unimportant and they retain large fractions of their initial masses .
clumps that are heated with @xmath228 evolve quasi - statically , steadily adjusting themselves to the pressure of the heated pdr shell .
the evolution of these clumps can be determined from simple steady - state equilibrium considerations . in their final configuration
, the clumps may retain a fraction of their initial cold gas , surrounded by a warm pdr envelope , in pressure equilibrium with the icm .
we consider two extremes in density profiles of the clump , a constant density clump and a truncated isothermal sphere ( @xmath229 ) in pressure equilibrium with the icm .
first , we solve for the evolution of a clump of constant density . from conservation of mass , @xmath230 where @xmath231 and @xmath232 are the number densities in the final cold clump core and the pdr , and @xmath233 , and @xmath234 are the radii of the initial clump , remnant clump , and the pdr envelope respectively . as the final configuration is a remnant clump in pressure equilibrium with the pdr and the icm , the final density of the core @xmath235 , and @xmath236 further , we know that the column density through the pdr envelope is @xmath83 , as the remnant clump is shielded from heating .
therefore , @xmath237 using equations ( [ ba ] ) and ( [ bc ] ) to eliminate @xmath232 and @xmath234 , equation ( [ bb ] ) can be written as @xmath238 where @xmath239 and @xmath240 . for complete photodissociation @xmath241 , or @xmath242 .
for the standard case @xmath171 , @xmath172 and @xmath120 , and for @xmath243 , equation ( [ etanupc ] ) can be used to define a critical column density for complete photoevaporation . @xmath244
all clumps with an initial column density lower than @xmath245 are thus eventually completely photodissociated .
alternately , a clump with an initial column density @xmath245 has to be exposed to an fuv field ( @xmath4 ) greater than or equal to that given by equation ( [ etacritpc ] ) to be completely photodissociated .
figure 7 shows a plot of the final column density of a clump , @xmath246 , as it is exposed to various fuv fields , measured by @xmath4 , for @xmath171 , @xmath172 and @xmath120 .
the figure shows trajectories followed by the clump column density as the fuv field is increased , as might happen when the clump slowly emerges into the pdr region from the shielded interior of the molecular cloud .
note that the abscissa is @xmath246 and not the initial column density @xmath247 .
the outer column of cold clump gas @xmath83 that is initially heated is thereby discounted , and as soon as the fuv field is `` turned - on '' at time @xmath188 , @xmath248 . as the fuv field ( @xmath4 ) increases , the clump begins to lose mass due to heating of the outer shell , and its column density begins to decrease with increasing @xmath4 . as @xmath4 approaches the critical value for complete photodissociation for that particular @xmath247 , the column density rapidly decreases with @xmath4 and the clump photoevaporates completely . if the final strength of the fuv field is less than that required for complete photoevaporation , the clump column density remains fixed at a point on its trajectory , as determined by the value of the parameter @xmath4 .
therefore , given a clump s initial column density and the strength of the local fuv field , the final column density of the clump can be determined .
equivalently , the observed @xmath246 and @xmath4 can be used to determine the initial column @xmath1 of the clump .
the total mass loss of the clump during its evolution can also be evaluated from equation ( [ etanupc ] ) .
figure 7 also shows the contours for the mass loss being 5% , 50% and 95% of the initial mass , for different values of @xmath246 and @xmath4 .
the evolution of a clump with an isothermal density profile can be determined analogously .
the initial density distribution in the clump is given by @xmath249 , where @xmath250 is the density at the clump surface , and the initial mass of the clump is @xmath251 .
we assume that the density in the heated pdr shell in the final configuration of the clump is constant . utilizing equations ( [ ba]-[bc ] ) for this case
, we obtain an equation very analogous to equation ( [ etanupc ] ) where now @xmath252 @xmath253 and @xmath254 . an effective average column density through the clump can be defined by the ratio of the mass to area of the clump , @xmath255 .
the evolution of a clump with an isothermal density profile is thus qualitatively similar to that of a constant density clump , but with a different effective column density for the same mass .
these solutions apply to clumps that are massive enough for gravity to be important in determining their density structure , and therefore have appreciable central density concentrations .
in the above analysis , simplifying approximations were made , which are now relaxed in a numerical code to obtain complete and more exact solutions ( see appendix d for details ) .
one of the major differences between the numerical code and most of the analytic equations is the inclusion of gravity in the numerical code .
we first discuss the results of a numerical simulation of a clump generated by turbulence and heated impulsively .
this case is hereafter referred to as case v ( for `` vacuum '' ) .
a vacuum boundary condition is used so that the clump is free to expand , even in the absence of any heating .
initially , the thermal , turbulent and magnetic pressures in the clump are all assumed to be equal ( @xmath256 ) , and the magnetic pressure scales with the density to the power @xmath120 . the fuv field is turned on instantly ( @xmath6=0 ) .
the clump is assumed to be a constant density sphere , with a column density given by @xmath257 , and in a fuv field characterized by @xmath258 in the outer layers .
the value of @xmath1 corresponds to that typical of clumps in clouds , for example , @xmath259 .
clumps exposed to fuv fields with @xmath260 are heated to @xmath261k on their surfaces ; if their shielded centres are @xmath262 , then the ratio of sound speeds @xmath263 .
these initial conditions imply the clump lies in the shock compression regime of figure 2 , and we expect from our analytic solutions that it should photoevaporate in less than one sound crossing timescale @xmath49 ( from eq . [ [ tpe2simtext ] ] ) , and @xmath264 . the density , pressure and temperature are scaled to the initial values in the clump and the velocity is in units of the clump thermal speed @xmath47 . ]
figure 8 shows the density , velocity , pressure and temperature profiles in the gas as a function of radius at three different instants of time , @xmath265 and @xmath266 .
as the outer layers heat up and the increased thermal pressure causes them to expand , a shock propagates into the clump , compressing the cold inner clump gas .
the shock strengthens as it progresses to the centre of the cloud , as evidenced by the increasing density and velocity at the position of the shock front .
the central cold gas gets compressed to very high densities , mainly due to the convergence of the radially - moving shocked gas . in figure 8
we only show the evolution prior to the shock reaching the centre of the cloud . after the shock reaches the centre , the compressed gas rebounds and the cold inner clump undergoes radial oscillations as it settles into pressure equilibrium with the warm , expanding pdr outflow in the outer layers . the mean density of the compressed clump gas increases with time , as expected from our analytical solutions ( eq . [
[ al ] ] ) .
as the clump loses mass , the fuv penetrates deeper and the clump is completely photoevaporated in @xmath267 .
our analytical solution ( eq . [ [ tpe2simtext ] ] ) predicts a photoevaporation timescale of @xmath268 , which agrees very well with the numerical solution .
the photoevaporation timescale derived for a clump with this column density , @xmath257 , and density , radius and sound speed given by @xmath269 respectively , is @xmath270 years . the compression of a clump by strong shocks due to fuv heating can raise the central densities by large factors .
this , however , did not render the clump gravitationally unstable to collapse in the above case because @xmath118 was chosen to be @xmath271 , which raises the magnetic pressure in dense gas sufficiently to prevent gravitational collapse ( chandrasekhar 1961 ) .
the scaling index of the magnetic pressure with density in clumps depends on the orientation of the magnetic field in the clump , and on whether the field is frozen into the neutral gas . for a frozen , uniform , unidirectional magnetic field , @xmath272 .
the magnetic field configuration in clumps is considerably more complicated ( ward - thompson et al .
2000 ) , and ambipolar diffusion may allow the neutral gas to slip past magnetic flux lines ; making it difficult to define @xmath118 unambiguously .
we investigate the possibility of triggering gravitational collapse in clumps through a simulation where we adopt @xmath273 , and all other parameters identical to those in case v. such a clump is expected to undergo gravitational collapse based on our earlier analysis ( eq .
[ [ collapse1e6 ] ] where @xmath257 , and figure 5 ) . .
] the evolution of the clump is initially similar to that in case v. the outer surface expands , and a shock is driven into the clump centre . as the shock reaches the centre , the density of the cold clump gas increases and the clump radius decreases , until the central region becomes gravitationally unstable .
a collapsing `` core '' is thus formed .
this can be seen in figure 9 , which shows the mass of the clump as a function of time .
the mass of the collapsing core is also shown in the figure , along with the results of case v. the core mass is seen to increase above that contained in the clump just before collapse as the core accretes mass from the warm pdr gas . a significant fraction ( @xmath274 ) of the initial clump mass is driven gravitationally unstable to collapse , on very short timescales . in the absence of sufficient magnetic field support , exposure to strong fuv fields can thus trigger star formation in clumps .
we also conducted a simulation with a weaker fuv field ( @xmath180 ) , and found that the central regions collapsed in that case too , but the collapsing core contained only about @xmath275 of the initial clump mass . although these triggered collapse solutions are interesting and instructive
, we emphasize that for initially stable clumps with @xmath276 , they hold only for @xmath277 , which requires ambipolar diffusion .
since ambipolar diffusion timescales are longer than the shock compression timescale , it is likely that realistic clumps require @xmath278 and that triggered collapse does not often occur .
pressure confined clumps are modelled as bonnor - ebert spheres which are in hydrostatic equilibrium .
we construct a bonnor - ebert sphere by setting up a density distribution for a clump with thermal , magnetic and turbulent pressure support and in hydrostatic equilibrium . the density in the clump decreases as @xmath280 in the outer regions ( similar to an isothermal sphere distribution ) , and flattens out in the central regions .
the ratio of central to surface density @xmath281 , determines the stability of the sphere against gravitational collapse , with a value exceeding 14.4 being `` critically unstable '' for an isothermal , non - magnetic gas . in these runs with a confining icm ,
the outer 200 zones are used to represent the icm .
the icm has the same pressure as the surface of the clump and is a constant density medium with a density ( and temperature ) contrast of 1000 .
as discussed in 4 , for clumps confined by an icm , a slow turn - on of the fuv field may be more appropriate .
the fuv field in the numerical run is turned on in a time @xmath282 , and we increase the temperature on this timescale so that the heated outer shell reaches its maximum temperature at @xmath283 .
figure 10 shows the results for a case with varying @xmath281 and where @xmath284 , and @xmath256 . for bonnor - ebert spheres with @xmath285 , the outer region with an isothermal density profile contains more than half the mass , and the clump evolution resembles that of an isothermal sphere . in the figure ,
the mass of the clump is shown with time along with the analytical results for an isothermal sphere and a constant density sphere with the same initial column .
the numerical results for three values of the central to surface density ratios , @xmath286 and @xmath287 are indicated .
for central concentrations greater than that represented by @xmath288 , the analytical results for an isothermal pressure - confined sphere match the numerical results , and the agreement increases as @xmath281 increases and the bonnor - ebert clump configuration approaches a @xmath289 density profile . for @xmath290 ,
the clump is a constant density sphere , and the corresponding analytical calculations apply .
a slight discrepancy can be noted in the figure between the two results , and this is due to the differences in the more realistic numerical model , i.e. an exponential drop in the pdr shell temperature with column and the inclusion of gravity . for intermediate central density concentrations , as represented by the case with @xmath291 , the mass loss rate is in between the two analytical extremes , as expected . and @xmath287 .
in this case , @xmath292 , and @xmath293 .
the corresponding analytical solutions for a constant density sphere ( dotted line ) and an isothermal sphere ( dashed line ) are also shown for comparison . ]
spherically asymmetric mass flows from photoevaporating clumps can cause a rocket effect on the clump , as first noted by oort & spitzer ( 1955 ) in their theory on cloud acceleration by ionizing radiation from ob stars .
clumps exposed to fuv radiation in pdrs see a stronger incident flux of photons from the direction of the ob stars , though a significant flux due to backscattered radiation from dust on the opposite surface tends to reduce this asymmetry ( see 5.2 ) .
the front and back hemispheres of the clump ( with respect to the ob star ) are thus heated to different temperatures ( different @xmath4 ) and the escaping material has a higher flow velocity on the star - facing surface of the clump .
the net thrust on the clump accelerates it away from the star , and is known as the rocket effect . with the help of our analytical expressions for the evolution of an impulsively - heated , photoevaporating clump in a pdr , we now evaluate a simple criterion which determines the clump size at which the rocket effect becomes significant . a simple 1-dimensional formulation for the acceleration due to the rocket effect is given by @xmath294 where @xmath295 and @xmath296 are the flow velocities at the front and back surfaces respectively and are assumed equal to the pdr sound speed @xmath32 at those surfaces .
we first compute the acceleration due to the flow from each hemisphere separately , take the difference to estimate the net acceleration , and then calculate the net velocity @xmath297 attained by the clump . the mass loss from a hemisphere
is given by @xmath298 , where @xmath299 is the density at the base of the flow .
taking only the component of the momentum in the flow along the direction to the star , we obtain @xmath300 the net acceleration on the clump is therefore given by @xmath301\ ] ] where the suffixes @xmath302 and @xmath303 denote the quantities for the front and back surfaces of the clump , and the clump radius @xmath304 is determined by the dominant flux on the front surface .
as the initial expansion / compression phases in the evolution of a photoevaporating clump are typically short ( see 5 ) , we ignore this initial phase of evolution and assume the clump is evolving such that its column density is at its critical value @xmath305 , for the standard values of @xmath306 and @xmath118 ( see eq .
[ [ aetext ] ] ) . along with the relation for the column through the flow @xmath307 for each surface and the mass of the clump @xmath308 ,
( see appendix , eq .
[ [ aa ] ] ) , we thus obtain for the net acceleration of the clump , @xmath309 \label{rocket}\ ] ] the clump will remain in the pdr of the gmc and be subject to photoevaporation if the rocket velocity @xmath297 attained in a time @xmath310 is less than @xmath311 , or equivalently if @xmath312 x_{pdr } = r_r \label{rocketsize}\ ] ] there is thus a critical size @xmath313 of clumps , where for @xmath314 they get rocketed back into the shielded molecular cloud , and for @xmath315 remain in the pdr and get photoevaporated or survive to photoevaporate in the region . for example ,
for front and back surface flow velocities of 3 km s@xmath15 and 2 km s@xmath15 respectively , @xmath316 km s@xmath15 and @xmath317 km s@xmath15 equation ( [ rocketsize ] ) gives a critical clump size @xmath318 .
clumps much smaller than this will get rocketed back into the molecular cloud as soon as they enter the pdr , partially evaporating in the process .
we emphasize again that the rocket effect is significant only for turbulent clumps which undergo impulsive heating and photoevaporative flows .
pressure - confined clumps are gradually heated on their exteriors and do not go through phases of evolution with rapid mass outflows , and hence experience an insignificant rocket effect .
the mass lost by a clump before it gets accelerated out of the pdr can be estimated by integrating equation ( [ momgain ] ) with time , where only the components of velocity and mass flow along the direction towards the source are to be considered .
thus the mass of the clump is approximately given by @xmath319 where @xmath320 is the difference in the flow velocities on the two sides of the clump .
for typical values of @xmath321 km s@xmath15 , and @xmath322 km s@xmath15 , in the time it takes for @xmath297 to reach @xmath311 , the clump loses about @xmath323% of its initial mass .
therefore clumps which are rocketed back into the cold molecular cloud interior also lose significant fractions of their mass before they attain high enough velocities to keep up with the advancing ionization front .
we begin our discussion by summarizing our results for both the case where clumps are produced by turbulence in regions with insignificant interclump pressure and the case where clumps are initially stable structures confined by interclump pressure .
consider first the scenario where clumps are continually generated by turbulence in the molecular clouds and the pdr .
initially , the evolution of the clump is determined by its column and the strength of the fuv flux .
if @xmath324 , or @xmath325 for @xmath171 , @xmath172 and @xmath326 , there is a very brief period of shock compression but the shock quickly dissipates before making it to the clump centre and the clump radius shrinks only slightly .
subsequently , the clump expands until @xmath327 , at which point the photoevaporative flow halts the expansion and proceeds to shrink and compress the clump gas , maintaining @xmath153 , or a column @xmath328 . if @xmath329 , then the shock propagates to the centre of the clump , compressing it significantly , followed by an evolution at @xmath330 identical to that above . during the final @xmath153 evolution of the shrinking , compressing ,
evaporating clump , the rocket effect may play a significant role depending on the physical size of the clump at this time .
if @xmath331 , the clumps will be accelerated into the molecular cloud , losing a significant fraction of their mass in the process . if @xmath332 , the clumps will tend to evaporate in the pdr region . however , because in the turbulent model clumps are forming in the pdr , including the region near the if , some of these clumps will survive to enter an advancing region . if @xmath333 and an region is advancing , then most of these large clumps will survive to be engulfed by the region .
the lifetimes of turbulent clumps are of order several sound crossing times of the initial size of the clump ( eq.[[tpe2simtext ] ] and eq.[[az ] ] ) . as discussed previously ,
this is somewhat longer than clump lifetimes in the absence of an fuv field .
since turbulent clumps are being continuously formed , the steady - state abundance of clumps could actually increase in the presence of an fuv field .
based on these results , the following predictions can be made for the turbulent scenario of photoevaporating clumps in pdrs . 1
. there will be an enhanced population of clumps with columns @xmath334 .
note that this implies that in comparing this population from one pdr to another , the pdr with the lower @xmath4 ( e.g. , due to lower @xmath13 ) will have clumps with smaller column densities .
2 . there will be a population of small clumps ( @xmath335 ) moving with velocity @xmath336 into the molecular cloud at @xmath337 .
3 . compared with clumps in fuv - shielded regions , the clumps in turbulent pdrs will be smaller , denser and potentially more numerous .
4 . similarly , the clumps entering the regions will be smaller and denser than the molecular cloud clumps , and will have columns peaked at @xmath338 .
we note here that in the present analysis , turbulent clumps are geometrically idealized as constant density spheres . in reality , turbulently generated structures may also be sheet - like or filamentary in nature .
let us consider two simple representations of these geometries , disks and cylinders .
for these structures to survive immediate photodissociation , the column density through their shortest dimension has to be greater than @xmath83 .
disks with larger column densities through their midplane establish photoevaporative flows in an fuv radiation field . when the outer boundary of the heated gas reaches a distance equal to the radius of the disk , the flow diverges and the density in the flow ( refer eq.[[c ] ] and eq.[[aa ] ] ) begins to drop as @xmath339 if @xmath340 . therefore , the mass loss rate from the disk is essentially similar to that of a spherical clump of size equal to the disk radius ( johnstone et al .
1998 ) photoevaporative flows from a cylindrical structure differ slightly from those off a spherical clump of the same radius .
the density at the base of the flow is now @xmath341 and therefore , the critical column density @xmath342 . the mass loss rate from the cylindrical clump ( following shock compression or initial expansion ) no longer depends on the radius and is a constant given by @xmath343 where @xmath344 is the length of the cylindrical filament .
eq.[[ao ] ] ) . in summary , for turbulent clumps which are likely to be sheets or filaments , the mass loss from a thin sheet or disk of dimension @xmath345 is best modelled as a clump of radius @xmath345 . the mass loss rate from a disk of radius @xmath345 and thickness @xmath346 is similar to that of a sphere of radius @xmath345 and is @xmath347 , but the evaporation timescale is @xmath348 and hence much smaller than the sphere ( @xmath349 ) for @xmath350 .
filaments of radius @xmath345 evolve somewhat differently than spheres of radius @xmath345 ; their mass loss rates do not decline as @xmath345 shrinks .
however , the evaporation timescales evolve similarly for cylinders or spheres since for spheres @xmath351 , whereas for cylinders @xmath352 .
consider next the scenario where clumps are pressure - confined structures which are bounded by a low - density icm , and the fuv turn - on timescale is of order the crossing time @xmath353 for clumps through the pdr .
for small clumps with @xmath354 , the fuv turn - on time @xmath228 , which means that the fuv quasi - statically heats and expands an outer pdr surface on the clump , continually maintaining a pressure @xmath355 in the pdr region and in the cold clump region .
again , the evolution of the clump is determined by its column and the strength of the fuv flux . if @xmath356 ( for @xmath357 , and @xmath358 ) , the clump is completely heated and photodissociated as it enters the pdr from the molecular cloud .
if @xmath359 , the clump shrinks to a smaller cold clump surrounded by a pdr shell . since the evolution is quasi - static , there is insignificant rocket effect on these clumps . in this scenario , small cold clumps , protected by pdr shells ,
can survive passage through the pdr and into the region .
however , the minimum initial column for such a clump is @xmath360 ( for @xmath121 , and @xmath358 ) . for large clumps , @xmath361 , @xmath362 , which means that shocks compress the clump and strong photoevaporative outflows
are initiated as the clump enters the pdr .
clumps with @xmath363/3 , are compressed by shocks which propagate to the centre , but they finally relax to an analogous situation as above , i.e. , a thick pdr shell surrounding a small mass cold clump if @xmath359 and totally photodissociated otherwise .
clumps with @xmath364 have a brief period of shock compression and pdr expansion , but rapidly relax to a configuration with a thin pdr shell and most of the initial mass still in the cold shielded core .
all the large clumps experience a rocket effect .
however , as shown earlier , large clumps with @xmath315 do not experience enough rocket acceleration to prevent them from entering the region .
therefore , in the icm confined scenario , a variety of clump sizes enter the region .
however , the small clumps entering the region were initially clumps of larger size and mass in the molecular cloud , and they enter the region with large protective pdr shells . one can compare the abundance of clumps of various sizes in pdrs with the abundance of clumps in the molecular cloud , under the assumption that icm - confined clumps are long - lived ( the formation time of clumps exceeds the crossing time @xmath353 ) .
small cold clumps are converted to pdrs but larger cold clumps are converted to small cold clumps with pdr shells and this partially replaces the small clump population .
thus , there will be a significant drop in the cold clump population with @xmath365 .
however , the population @xmath366 is little affected .
therefore , a variety of clump sizes enter the region , but there is a suppression of clumps with columns @xmath365 compared to the distribution in the shielded molecular cloud . based on these results ,
the following predictions can be made for clumps initially confined by interclump pressure . 1 .
compared with small cold clumps ( @xmath365 ) in similar pressure regions of the molecular cloud , clumps of the same column density in pdrs will have a smaller relative population .
correspondingly , small and large clumps enter the advancing regions , but the relative number of small clumps is suppressed .
larger clumps are relatively unaffected , but do experience an initial transient period of shock compression and photoevaporative flow .
3 . small clumps ( @xmath365 ) in the fuv - illuminated region will have small masses relative to their pdr shells .
[ [ section ] ] we have shown that small clumps tend to be destroyed before they are overtaken by the ionization front of the advancing region .
this would affect the number , size distribution , and structure of pressure - confined clumps that enter an region , and therefore affect the evolution of the region ( see bertoldi 1989 and bertoldi & mckee 1990 for a discussion of the effect of photoevaporating clumps on the evolution of an region ) .
clumps that survive to enter the region will thus have undergone significant evolution from their initial states .
these clumps could be considerably denser if they are still vigorously photoevaporating as they enter the region , where they continue to photoevaporate ( bertoldi 1989 , bertoldi & mckee 1990 ) .
we will discuss the propagation of clumps into regions in a separate paper , with application to the eagle nebula .
we conclude our discussion by using our results to critique the previous inferences of dense ( @xmath367 @xmath20 ) fuv heated clumps in pdrs and to propose several observational consequences of our results .
our photoevaporation models provide arguments against the interpretation that dense ( @xmath368 @xmath20 ) pdr surfaces of clumps give rise to the observed high excitation co lines .
the fundamental problem is the propagation of the fuv photons to regions of such high density .
if densities of @xmath369 @xmath20 are in fact heated to @xmath370k , then the thermal pressures in these regions are much larger than the pressures in the interclump gas .
the only way this is possible is if : ( i ) the clumps are gravitationally bound or ( ii ) the warm surface regions are photoevaporating , as discussed in this paper . in either case
, the density in the clump will smoothly fall off with radius , @xmath371 where @xmath372 , until the transition to the icm . however , fuv photons which heat the co are only able to penetrate through a column @xmath373 @xmath3 , where @xmath374 this equation reveals the size @xmath345 of the fuv heated region if a density @xmath375 is required in this region .
it implies that for the fuv to heat gas at densities @xmath376 @xmath20 , the clumps must be very small with size scales of @xmath377 @xmath378 cm at this density .
gravitationally bound clumps have densities @xmath368 @xmath20 at sizes @xmath379 cm ( e.g. , shu 1977 ) . in such clumps , the fuv would not penetrate to such high densities , but would be absorbed in the lower density regions further out .
such small scales as @xmath380 cm also argue against unbound clumps explaining the high temperature co emission .
we have shown in this paper that such a small clump will either be rocketed out of the pdr or photoevaporate in roughly a sound crossing time , or @xmath381 years , a time so short as to rule out this possibility . the size scale of 100 au is suggestive of disks around young stars rather than clumps , so that one might appeal to photoevaporating disks like the `` proplyds '' in orion ( e.g. , johnstone et al . 1998 ) . here
, the enormous mass reservoir of the disk greatly lengthens the photoevaporative timescale .
however , the area filling factor of these photoevaporating disks required to match the co high - j line intensities in , for example , the orion bar region is of order @xmath42 ( burton et al .
such a high area filling factor requires a volume density of @xmath382 stars pc@xmath383 , more than an order of magnitude higher than that observed in the densest region of the stellar cluster near the trapezium ( mccaughrean et al . 1994 ) .
photoevaporating proplyds are thereby also ruled out .
therefore , it seems quite unlikely that fuv radiation is heating dense , @xmath368 @xmath20 , pdr surfaces of clumps or that photoevaporating proplyds can explain the mid - j co emission .
we suspect that the co emission arises in @xmath384 @xmath20 surfaces of large ( @xmath385 cm ) clumps , and that the chemistry , heating and/or dynamics needs to be modified in pdr models in order to better match the observations .
there are several observational consequences implicit in the scenario of a strongly photoevaporating ( i.e. , @xmath123 ) clump .
first , the molecular material from the clump core is effectively advected out into the pdr region , as has been discussed by bertoldi & draine ( 1996 ) , strzer & hollenbach ( 1998 ) , hollenbach & tielens ( 1999 ) .
this can have strong effects on the chemistry , primarily because h@xmath27 can exist in surface regions where it would have been atomic in a stationary , steady - state case . in turn
, the higher abundance of h@xmath27 can chemically enhance the abundance of other minor species , such as h@xmath386 ( bertoldi & draine 1996 ) and co and can modify the heating and cooling processes and therefore the pdr temperature
. initial results ( strzer & hollenbach 1998 and strzer 2000 , private communication ) suggest that advection will enhance the intensities of the mid - j co lines , by creating more warm co near the surface . finally , the flow will broaden line widths to values @xmath387 .
however , it should be noted that many common molecules , such as co , hco@xmath388 , cs , hcn , cn still exist primarily at column depths @xmath389 from the surface , where the advection velocities are low .
broader linewidths should be seen in ( 158@xmath390),(63@xmath390 ) and possibly the h@xmath27 ( 2@xmath390 ) lines .
we compare our results with available observational data on clump characteristics from some well - studied star - forming regions , the orion bar , m17sw , ngc 2023 and the pdr surrounding the rosette nebula .
clump column densities ( @xmath391 ) can be estimated from measured or inferred densities and sizes of clumps in these pdrs .
densities are usually inferred indirectly from comparing observed line intensities and line ratios with available pdr chemical models in most of these cases .
clump sizes are ill - determined , as they are usually too small to be resolved directly and only upper limits to the sizes are available , with the pdrs in rosette and ngc 2023 being possible exceptions .
the fuv field strengths in these pdrs can be estimated more accurately , as the luminosities of the illuminating o / b stars is reasonably well - determined .
we use these field strengths and the dust temperature models of hollenbach , takahashi & tielens ( 1991 ) to estimate the temperature of dust in the shielded clump interior . as the interior gas temperature
is closely coupled to the dust temperature , we set the two equal and thus determine the sound speed @xmath47 in the cold clump gas .
the temperature to which the fuv field heats the surface clump gas is determined from the pdr models of kaufman et al .
( 1999 ) , and thus @xmath32 is obtained , allowing us to estimate the value of the parameter @xmath4 in each region .
below we first determine @xmath391 and @xmath4 for each observed region , and then use these parameters , compared to figures 2 and 7 , to infer their nature and evolution .
[ [ orion - bar ] ] orion bar + + + + + + + + + the orion bar has been observed in many atomic and molecular line transitions in the sub - millimeter and infrared wavelength regions ( e.g. van der werf et al .
1996 , tauber et al .
1994 , young owl et al . 2000 ,
marconi et al . 1998 ) .
the molecular line emission peaks at about 0.03 pc from the if , and the gas between this dense ridge and the if is mainly neutral atomic hydrogen .
the observed line intensities and line ratios appear to be best explained by a two - component model for the pdr gas , with small ( @xmath392 pc ) dense ( @xmath393 @xmath20 ) clumps , required to match the line emission and an interclump medium @xmath394 @xmath20 which causes the chemical stratification of the edge - on pdr .
the local fuv field is estimated as @xmath395 .
such an fuv field incident on clumps with these inferred densities would heat their surfaces to temperatures @xmath396 k ( kaufman et al . 1999 ) .
the gas temperature in the interior of clumps exposed to this fuv flux is expected to be @xmath397 k ( hollenbach et al . 1991 ) .
we thus obtain @xmath398 and @xmath399 .
[ [ m17-sw ] ] m17 sw + + + + + + multilevel molecular line observations in cs ( wang et al .
1993 , evans et al .
1987 ) and in several fine structure lines ( , , , meixner et al .
1992 ) indicate that the m17 sw cloud core consists of numerous high density clumps with densities @xmath400 @xmath20 .
maps of co@xmath388 emission compared with theoretical models suggest that the emission is produced in the warm surface layers of pdrs in dense clumps ( strzer et al .
1995 ) of sizes @xmath401 pc . the local fuv field near the region / molecular cloud interface is @xmath402 , which implies pdr temperatures of about 2000k on the clump surfaces and cold clump gas temperatures @xmath403k , therefore , @xmath404 .
the column densities of clumps are typically @xmath405 , as estimated from typical densities and sizes ( meixner et al .
1992 , strzer et al .
1995 ) .
[ [ ngc-2023 ] ] ngc 2023 + + + + + + + + the well - studied pdr in the reflection nebula ngc 2023 , at a distance of 475 pc , is illuminated by a b star and is well - studied ( e.g. , wyrowski et al .
2000 , steiman - cameron et al .
1997 , howe et al .
observations of atomic fine structure lines indicate the presence of clumpy gas in the pdr with densities @xmath406 @xmath20 , and an icm density of @xmath407 @xmath20 ( steiman- cameron et al .
the pdr is situated about 0.1 - 0.2 pc away from the exciting star , and has a width @xmath408 pc .
the clumps have sizes @xmath409 pc ( wyrowski et al . 2000 ) , and therefore @xmath410 . the fuv field strength has been estimated at @xmath411 , and from the pdr models of kaufman et al .
( 1999 ) , @xmath412 km s@xmath383 . with a clump gas temperature of @xmath413 k for this value of @xmath13
, we obtain @xmath414 .
it should be noted that the level of turbulent energy in the ngc 2023 pdr as estimated from observed linewidths is low ( wyrowski et al .
2000 ) , which suggests that clumps are not rapidly re - formed by turbulence in the pdr .
[ [ rosette - pdr ] ] rosette pdr + + + + + + + + + + + the rosette nebula surrounding the young open cluster ngc 2244 , is an expanding region at a distance of 1.6 kpc .
the region / molecular cloud interface shows a distinct ridge of emission , with substructure on small scales down to 0.5 pc ( schneider et al .
the pdr region is very extended , and the fuv photons penetrate deep into the cloud with an interclump medium density @xmath415 @xmath20 ( blitz 1991 , williams et al .
the icm near the edge of the hii region may even by partially ionized gas as suggested by the pervasive h@xmath112 emission observed ( block et al .
this region has a low fuv field , with @xmath416 about 15 pc away from the ob star cluster , and dropping to lower values of 10 - 50 , about 30 pc away from the cluster ( schneider et al .
clump densities required to match observed cii line intensities are @xmath417 @xmath20 .
cold clump gas temperatures are calculated to range from @xmath418 k. the rosette pdr thus has @xmath419 and @xmath420 , while further away from the region , @xmath421 .
[ [ section-1 ] ] using the above determined values of the parameters @xmath391 and @xmath4 for clumps in the above pdrs , we can locate them on the parameter plots of figures 2 and 7 .
figure 11 shows an overlay of our model results with inferred data for both the turbulent and pressure - confined clump models .
the data are consistent with both clump models and , within the limits of the uncertainties in arriving at `` observed '' @xmath391 and @xmath4 , agree reasonably well with our predictions .
turbulent clumps are expected to have columns fairly close to the critical value @xmath147 which is @xmath422 .
the observed data lie near the @xmath423 line , indicating that the observed clumps may be turbulent in origin and undergoing photoevaporation in the fuv field .
the exception is ngc 2023 .
the clumps in this pdr are observed to have negligible turbulence , and are thus not expected to have a column equal to @xmath147 .
the data are also consistent with clumps being pressure - confined . in such a situation
, we would not expect to see clumps beyond the 95% mass loss line , as seen in the figure .
however , there does appear to be a weak trend of increasing @xmath391 with @xmath4 , which if real is not easily explained in this model .
the clumps in ngc 2023 are probably pressure - confined , and seem to have lost almost 75% of their original mass through fuv heating .
high resolution observations of clumps in pdrs , with more accurate determinations of the sizes and densities of clumps would help in locating them more precisely on the @xmath154 plots , and thus be able to distinguish between the pressure - confined and turbulent clump models .
we have studied the effects of fuv radiation from young ob stars on the evolution of dense structures or clumps in photodissociation regions .
we find that the ambient fuv field penetrates through the surface of a dense clump and heats this surface layer to high temperatures , causing mass loss and thereby inducing photoevaporation of the clump . through analytic approximations and numerical hydrodynamical calculations , we determined the evolution and lifetimes of clumps , subject to various physical conditions and initial parameters .
the evolution of a clump is mainly determined by three parameters , the ratio of the initial column density to the column penetrated by the fuv flux , @xmath1 ; the strength of the fuv field , which we denote by the ratio of the sound speeds in the heated gas and the cold clump material , @xmath4 ; and the timescale for the turn - on of the fuv field , relative to the sound crossing time through the clump , @xmath424 .
we consider the evolution of turbulent , impulsively - heated clumps and also pressure - confined clumps in a pdr .
impulsively heated clumps with a photoevaporation parameter @xmath425 are initially compressed by a shock driven by the external high - pressure fuv - heated layer to high densities , and later lose mass as the fuv gradually penetrates through the entire clump .
clumps with @xmath324 go through an initial expansion phase after the rapid decay of the shock wave .
both clumps with @xmath329 and @xmath324 evolve toward the @xmath426 condition .
the final evolution toward complete photodissociation and heating occurs at constant column through the clump , the density increasing as the clump shrinks .
photoevaporation timescales in all these cases are typically a few @xmath49 , suggesting clump destruction times of @xmath427 years , under typical pdr conditions .
slowly heated clumps evolve quasi - statically , and if confined by an external icm pressure , a fraction of the initial cold clump mass may be retained if @xmath428 .
clumps with initial column densities @xmath429 are completely photoevaporated by the fuv flux .
we predict that in the turbulent scenario , observed clumps should all have columns close to the critical value for the local fuv field , @xmath430 .
clumps lifetimes are prolonged by photoevaporation and we also expect a higher steady - state abundance of clumps in pdrs as compared to the shielded molecular cloud interior . for pressure - confined clumps , there will be a decrease in the number of smaller clumps in pdrs , and many clumps will lose substantial fractions of their mass to extended warm pdr shells around them .
we compared our results to observations of some well - studied pdrs , such as the orion bar , m17sw , ngc 2023 and the rosette pdr , and find that the data are consistent with both clump models , but perhaps favour the turbulent clump interpretation .
clumps entering the region around an o / b star with a d - type ionization front are expected to have thus undergone significant evolution as they pass through the pdr .
we acknowledge support from the nasa astrophysical theory program under rtop 344 - 04 - 10 - 02 , which supports the center for star formation studies .
u. gorti was supported by a national research council associateship .
we thank k. r. bell , m. kaufman , c. mckee and h. strzer for useful discussions .
we first define several characteristic timescales during evolution .
the initial sound crossing time through the clump is denoted by @xmath432 .
we can similarly define a sound crossing time through the warm pdr layer , as @xmath433 .
finally , we define the timescale for the pdr layer to expand to a thickness comparable to @xmath132 , @xmath434 .
we relate the density @xmath435 at the base of the pdr shell ( @xmath436 ) to @xmath83 for epochs when the pdr shell has expanded to a thickness greater than the clump radius ( @xmath437 ) .
the fuv - heated shell of initial thickness @xmath438 expands as an isothermal flow into vacuum and the characteristics of the flow regions are essentially similar to that of the parker wind solution with no gravity ( parker 1958 ) . for simplicity
, it is assumed here that once the warm pdr region has expanded to @xmath439 or @xmath440 , the gas leaves the clump surface at @xmath32 and the velocity of the heated gas increases linearly with radius .
based on our hydrodynamical code results , this is a reasonable approximation to within about two clump radii , where most of the column density in the pdr gas lies .
the clump mass loss rate is determined by the penetration depth of the fuv flux , and hence @xmath83 . for a power law density profile in the shell ( @xmath441 ) ,
most of this column arises from the base of the shell , or , more precisely , from the region between the base ( clump surface ) @xmath442 and @xmath443 . from the equation of continuity and @xmath444 , the steady flow density profile of the warm pdr shell
is obtained as , @xmath445 valid for @xmath446 once the shell has expanded to at least two clump radii ( @xmath437 ) .
equation ( [ c ] ) can be easily integrated over @xmath345 to obtain the column through the warm pdr at any given instant of time .
let @xmath442 denote the instantaneous cold clump radius and @xmath447 the total radius of cold clump and warm pdr gas . at @xmath188 , @xmath448 .
using the fact that @xmath449 for @xmath437 , and as the fuv flux always penetrates through a column @xmath83 , we have @xmath450 we next write down the equations of momentum flux or pressure .
the pressure in the shocked cold clump gas is the sum of thermal , turbulent and magnetic pressures and given by @xmath451 where @xmath452 is the shocked clump density at time @xmath346 , @xmath133 is the initial clump density , @xmath453 is the initial ratio of turbulent and thermal pressures , and @xmath117 is the initial ratio of magnetic to thermal pressure in the clump .
the pressure at the base of the pdr flow , @xmath454 , is given by @xmath455 where @xmath456 is the outward flow at the base .
we assume that @xmath457 , and from equations ( [ aa ] ) and ( [ ac ] ) , @xmath458 the boundary between the `` shock compression '' and `` initial expansion '' regions of figure 2 is determined by comparing equations ( [ ab ] ) and ( [ ad ] ) .
values of the parameters @xmath1 and @xmath4 on this boundary are such that the shock dies out just before making it to the centre .
the pressure in the clump at this instant is about the same as the initial pressure and @xmath459 . in a time @xmath460 ,
the outer edge of the pdr shell expands out to @xmath461 , and the initial pdr column begins to decrease appreciably due to spherical divergence , allowing further fuv flux penetration . if @xmath462 , the pressure in the pdr flow is sufficiently high for the shock wave to reach the centre of the clump , compressing the clump and raising its central pressure so that @xmath463 .
such clumps lie in the `` shock compression '' of figure 2 .
if the initial parameters are such that @xmath464 , the shock wave rapidly dies out before reaching the centre and the clump then stops contracting and begins to expand at a speed @xmath465 , maintaining pressure equilibrium .
these two evolutionary sequences are separated by the ( @xmath466 ) boundary obtained setting @xmath467 , or from equations ( [ ab ] ) and ( [ ad ] ) , @xmath468 for values of @xmath1 and @xmath4 typical in pdrs such as orion ( @xmath469 ) this relation can be approximated as @xmath470 .
a photoevaporation parameter @xmath471 can be defined such that the relation @xmath153 demarcates the two regions .
clumps with @xmath329 are shock compressed and those with @xmath324 initially expand .
as the flow expands in a shock compressed clump , the clump shrinks by losing mass at a rate @xmath472 initially , a shock is driven into the clump , but we do not attempt to analytically model the propagation of the shock wave in detail .
we assume that the shock travels inward at the pdr sound speed and the entire clump gets compressed in a time @xmath473 , where @xmath474 .
we first solve for the clump radius , mass and mass loss rate for @xmath475 , during the epoch when the shock propagates to the centre .
we assume that the clump radius decreases with a constant speed @xmath476 so that @xmath477 the mass of the cold clump decreases with time , and at @xmath185 is given by @xmath478 where @xmath479 . from equations ( [ aa ] ) , ( [ af ] ) , ( [ ag ] ) and ( [ ah ] ) , we have @xmath480 we solve for @xmath186 and thereby calculate @xmath481 . to do this
we need to calculate the clump compression at @xmath185 .
the clump gets compressed by the shock to a new density @xmath482 which can be determined from the condition of pressure equilibrium at the clump surface .
the momentum flux conservation equation across the `` front '' which marks the boundary between clump and fuv heated flow is then @xmath483 in moderate to strong fuv fields , the pdr sound speeds and hence the shock velocities ( assumed equal to @xmath32 ) are high , and the resulting compression is also high . as the magnetic pressure @xmath109 , scales with a higher power of the density than the thermal pressure @xmath111 , ( we use @xmath484 ) and as the initial magnetic pressure is comparable to the thermal pressure , the thermal contribution to the pressure in the compressed gas can be ignored . in the expanded flow ,
the density is low and here the thermal pressure and dynamical pressure dominate .
equation ( [ aj ] ) can be simplified to give the density of the compressed gas @xmath485 from equations ( [ aa ] ) and ( [ ak ] ) , @xmath486 the mass of the clump at time @xmath185 can also be expressed as @xmath487 from equations ( [ ag ] ) , ( [ al ] ) and ( [ am ] ) @xmath488 at time @xmath185 , the radius of the shock - compressed clump is @xmath489 from equation ( [ ag ] ) . the unknown parameter @xmath186 is finally determined from equating the two expressions for the clump mass , equations ( [ ai ] ) and ( [ an ] ) . the evolution of the clump radius is thus known as a function of time for @xmath490 , from equations ( [ ag]),([al ] ) and the solution for @xmath186 .
these general equations are , however , fairly complex , and an approximate solution can be obtained for the case @xmath121 , and @xmath120 as , @xmath491 where @xmath492 .
we now solve for the evolution of the clump for @xmath170 .
after being compressed by the shock , the clump continues to lose mass through photoevaporation , with a mass loss rate , @xmath493 the mass of the clump at any given time ( @xmath170 ) can also be written as @xmath494 from equation ( [ ao ] ) and differentiating equation ( [ ap ] ) with time , the radius of the clump as a function of time is given @xmath495 a photoevaporation timescale @xmath189 for the clump can be defined as the time for the radius of the clump to shrink to zero , and from equation ( [ aq ] ) @xmath496 for @xmath121 , and @xmath120 , the above equations determining the evolution of the clump can be simplified as @xmath497 @xmath498 and @xmath499
clumps with large initial column densities and in low fuv fields quickly evolve to the point where their internal pressures are greater than that of the ( expanded ) thin heated surface layer .
they expand out into vacuum ( as they would even in the absence of an external fuv field ) at their sound speed @xmath47 for a time @xmath216 , until the pressure drops to that in the heated outer layer . at @xmath500
, there is pressure equilibrium and equation ( [ aj ] ) again holds .
the clump expands to a new radius @xmath501 and during expansion loses mass at a rate given by equation ( [ af ] ) .
the mass of the clump at @xmath500 is obtained as earlier , @xmath502 also , @xmath503 because there is significant expansion , @xmath504 , we assume that thermal pressure dominates in both the clump and the pdr flow . from equations ( [ aj ] ) and ( [ as ] ) @xmath505 equations ( [ ar ] ) and ( [ at ] ) can be solved for @xmath216 , and @xmath506 at times @xmath507 , clump expansion is halted , and the clump is now confined by the pressure at the base of the pdr flow
. the clump slowly loses mass , and shrinks to eventually become completely photoevaporated .
the time evolution of clump mass and size can be obtained by differentiating equation ( [ at ] ) which also holds for @xmath507 , with respect to time and equating this with the mass loss rate ( equation [ ao ] ) .
thus , @xmath508 @xmath509 the photoevaporation timescale is obtained by setting the clump radius to zero at @xmath189 to give @xmath510 and @xmath216 is given by equation ( [ te ] ) .
the fluid equations of motion for the system are solved using a 1-d spherical lagrangian hydrodynamics code .
the equations are solved using a finite difference method , with a numerical viscosity term , added as a pseudopressure , for accurate handling of shocks and discontinuities ( richtmyer & morton 1967 , bowers & wilson 1991 ) .
an isothermal equation of state is used throughout .
the accuracy and stability of the scheme was checked with standard test problems with known solutions .
the computational grid has 1000 equally spaced radial zones , except for the central @xmath511 , which was spaced logarithmically in radius .
this was done to increase spatial resolution at the centre , and it provides a more accurate calculation of any shock - induced collapse .
the fuv field is gradually turned on over a timescale , @xmath6 , which is varied to accommodate both the impulsive and slow heating cases described in 4 .
the temperature of an outer shell of material is thus raised continuously from @xmath512 to @xmath513 , reaching its maximum value of the pdr temperature at a time @xmath6 .
the temperature @xmath514 , exponentially drops off with the square of the column , @xmath515 , into the cloud , @xmath516 .
this profile was chosen to closely mimic the @xmath514 dependence with column predicted by pdr models for the values of @xmath13 and @xmath16 under consideration ( kaufman et al .
1999 ) .
allen , l. e. , burton , m.g .
, ryder , s.d .
et al . 1999 ,
mnras , 304 , 98 bakes , e.l.o . , & tielens , a.g.g.m .
1994 , apj , 427 , 822 bertoldi f. , 1989 , apj , 346 , 735 bertoldi f. , draine b. 1996 , apj , 458 , 222 bertoldi f. , mckee , c.f .
1990 , apj , 354 , 529 bertoldi f. , mckee , c.f .
1992 , apj , 395 , 140 blitz , l. & stark , a. 1986 , apjl , 300 , 89 blitz , l. 1991 , the physics of star formation and early stellar evolution , kluwer , p3 block , d.l . ,
dyson , j.e .
, & madsen , c. 1992 , , 390 , l13 burton , m.g . ,
hollenbach , d.j . , &
tielens , a.g.g.m .
1990 , apj , 365 , 620 chandrasekhar , s. 1961 , hydrodynamic and hydromagnetic stability , oxford : clarendon press crutcher , r.m . 1999 ,
apj , 520 , 706 dsert , f - x . , boulanger , f. , puget , j.l .
1990 , a&a , 237 , 215 weingartner , j.c . , & draine , b.t .
2001 , , 134 , 263 + elmegreen , b.g .
1999 , the physics and chemistry of the interstellar medium , p77 evans , n.j . ,
davis , j.h . , mundy , l.g .
, & vandenbout , p. 1987 , apj , 321 , 344 evans , n.j . 1999 , , 37 , 311 falgarone , e. & phillips , t.g . 1990 , apj , 359 , 344 genzel , r. , 1991 , galactic interstellar medium , saas - fee advanced course 21 ( heidelberg : springer - verlag ) goodman , a.a . , barranco , j.a . , wilner , d.j . , & heyer , m.h . 1998 , apj , 504 , 223 habing , h.j .
1968 , , 19 , 421 henyey , l.c . &
greenstein , j.l .
1941 , apj , 93 , 70 hill , j.k . & hollenbach , d.j .
1978 , apj , 225 , 390 hogerheijde , m.r . ,
janse , d.j . , van dishoeck , e.f . 1995 ,
a&a , 294,792 hollenbach d.j . , tielens a.g.g.m . 1999 ,
rev.mod.phys , 71,173 howe , j.e . , jaffe , d.t . , genzel , r. , & stacey , g.j .
1991 , apj , 373 , 158 hurwitz , m. , bowyer , s. , & martin , c. 1991 , apj , 372 , 167 jijina , j. , myers , p.c .
, & adams , f.c . 1999 ,
apjs , 125 , 161 johnstone d. , hollenbach d.j .
, bally j. 1998 , apj , 499 , 758 kaufman m.j . , wolfire m.g . , hollenbach d.j . , luhman m.l . 1999 ,
apj , 527 , 795 klein r.i . , mckee c.f .
, colella p. 1994 , apj , 420 , 213 kuchar , t.a .
, & bania , t.m .
1993 , apj , 414 , 664 luhman , k.l . ,
engelbracht , c.w .
, & luhman , m.l .
1998 , , 499 , 799l maloney , p. 1988 ,
apj , 334 , 761 marconi , a. , testi , l. , natta , a. , walmsley , c.m .
1998 , a&a , 330 , 453 mccaughrean , m.j . &
stauffer , j.r .
1994 , aj , 108 , 1382 mckee , c.f .
1999 , the origins of stars and planetary systems , eds .
lada , n.d.kylafis , kluwer , p29 meixner , m. , haas , m.r . , tielens , a.g.g.m . , erickson , e.f . & werner , m. 1992 , apj , 390 , 499 odell , c.r . ,
wen , z. , hu , x. 1993 , apj , 410 , 696 oort , j.h . , & spitzer , l.j .
1955 , apj , 121 , 6 parker , e.n . 1958 ,
apj , 128 , 664 plume , r. , jaffe , d.t . , evans , n.j .
1997 , apj , 476 , 730 scalo , j. 1990 , physical processes in fragmentation and star formation , kluwer , p151 schneider , n. , stutzki , j. , winnewisser , g. , block , d. 1998 , a&a , 335 , 1049 shu , f.h .
1977 , apj , 214 , 488 steiman - cameron , t.y . ,
scargle , j.d . ,
imamura , j.n .
& middleditch , j. 1997 , apj , 487 , 396 steiman - cameron , t.y . , hass , m. , tielens , a.g.g.m .
, & burton , m.g .
1997 , apj , 478 , 261 strzer h. , stutzki , j. & sternberg , a. 1995 , a&a , 296l , 9 strzer h. , hollenbach d.j . 1999 ,
apj , 515 , 669 strzer h. , hollenbach d.j .
1998 , apj , 495 , 853 stutzki , j. , stacey , g.j . , genzel , r. et al .
1988 , apj , 332 , 379 tauber , j. , tielens , a.g.g.m . , meixner , m. , goldsmith , p.f .
1994 , apj , 422 , 136 tielens a.g.g.m .
, hollenbach d.j 1985 , apj , 291 , 722 tielens a.g.g.m . ,
hollenbach d.j 1985b , apj , 291 , 747 van der werf , p.p . ,
stutzki , j. , sternberg , a. , krabbe , a. , 1996 , , 313,633 van der werf , p. , goss , w.m . ,
heiles , c. , crutcher , r.m . & troland , t.h .
1993 , apj , 411 , 247 wang , y. , jaffe , d.t . , evans , n.j . et al .
1993 , apj , 419 , 707 ward - thomson d. , kirk j.m .
, crutcher r. , greaves j.s . ,
holland w.s . , andre p. , 2000 ,
apjl , in press watson , w.d .
1972 , apj , 176 , 103 whitworth , a.p . 1979 ,
mnras , 186 , 59 williams , j.p . , blitz , l. , & stark , a.a .
1995 , apj , 451 , 252 williams , j.p . , blitz , l. , & mckee , c.f . 2000 ,
protostars and planets iv , p97 wyrowski , f. , walmsley , c.f . ,
goss , w.m . , &
tielens , a.g.g.m .
2000 , apj , 543 , 245 young owl , r.c .
, meixner , m. , wolfire , m. , tielens , a.g.g.m .
2000 , & tauber , j. apj , 540 , 886 | we present the results of an investigation of the effects of far ultraviolet ( fuv ) radiation ( @xmath0 ) from hot early type ob stars on clumps in star - forming molecular clouds .
clumps in fuv - illuminated regions ( or photodissociation regions or pdrs ) undergo external heating and photodissociation as they are exposed to the fuv field , resulting in a loss of cold , molecular clump mass as it is converted to warm atomic gas .
the heating , if rapid , creates strong photoevaporative mass flows off the clump surfaces , and drives shocks into the clumps , compressing them to high densities .
the clumps lose mass on relatively short timescales .
the evolution of an individual clump is found to be sensitive to three dimensionless parameters : @xmath1 , the ratio of the initial column density of the clump to the column @xmath2 @xmath3 of a warm fuv - heated surface region ; @xmath4 , the ratio of the sound speed in the heated surface to that in the cold clump material ; and @xmath5 , the ratio of the `` turn - on time '' @xmath6 of the heating flux on a clump to its initial sound crossing - time @xmath7 .
the evolution also depends on whether a confining interclump medium exists , or whether the interclump region has negligible pressure , as is the case for turbulence - generated clumps . in this paper , we use spherical 1-d numerical hydrodynamic models as well as approximate analytical models to study the dependence of clump photoevaporation on the physical parameters of the clump , and to derive the dynamical evolution , mass loss rates and photoevaporative timescales of a clump for a variety of astrophysical situations .
turbulent clumps evolve so that their column densities are equal to a critical value determined by the local fuv field , and typically have short photoevaporation timescales , @xmath8 years for a 1 m@xmath9 clump in a typical star - forming region ( @xmath10 ) .
clumps with insufficient magnetic pressure support , and in strong fuv fields may be driven to collapse by the compressional effect of converging shock waves .
we also estimate the rocket effect on photoevaporating clumps and find that it is significant only for the smallest clumps , with sizes much less than the extent of the pdr itself .
clumps that are confined by an interclump medium may either get completely photoevaporated , or may preserve a shielded core with a warm , dissociated , protective shell that absorbs the incident fuv flux .
we compare our results with observations of some well - studied pdrs : the orion bar , m17sw , ngc 2023 and the rosette nebula .
the data are consistent with both interpretations of clump origin , turbulence and pressure confinement , with a slight indication for favouring the turbulent model for clumps over pressure - confined clumps . |
confinement in mesons and baryons should be very similar since color dynamics is sensitive at large distances only to the net color charge of the interacting sources .
thus whether a quark is bound to an antiquark or to a diquark in a color @xmath1 ( as it must be for a baryon to be in an overall color singlet ) should make no difference at large separation ( or alternatively at high excitation ) . for similar reasons , the one - gluon - exchange forces in mesons and baryons
are very closely related : pairwise forces in baryons have exactly half the strength of those in mesons ( for identical spins and separations ) . given this close connection between meson and baryon dynamics and the success of the quark model in meson spectroscopy , it is surprising that there is still an unsettled qualitative problem in the quark model for baryons : the so - called baryon spin - orbit puzzle " that baryon spin - orbit splittings appear to be much smaller than expected from their one - gluon - exchange matrix elements @xcite .
however , by this criterion the mesons would also have a spin - orbit problem .
meson spin - orbit splittings are also much smaller than expected from their one - gluon - exchange matrix elements @xcite , but mesons have no spin - orbit problem because the normal " spin - orbit matrix element is largely cancelled by a strong inverted " spin - orbit matrix element from thomas precession in the confining potential .
the physics behind this cancellation has received support recently from analyses of heavy quarkonia , where both analytic techniques @xcite and numerical studies using lattice qcd @xcite have shown that the confining forces are spin - independent _ apart _ from the inevitable spin - orbit pseudoforce due to thomas precession . moreover , as has been known for more than ten years , the data on charmonia require an inverted spin - orbit matrix element from thomas precession in the confining potential to cancel part of the strength of the oge matrix element @xcite . if the charm quark were sufficiently massive , its low - lying spectrum would be rigorously dominated by one gluon exchange , and one indeed observes that the @xmath2 system is closer to this ideal .
conversely , as one moves from @xmath3 to lighter quarks , the @xmath4 wave functions move farther out into the confining potential and the relative strength of the thomas precession term grows .
it is thus very natural to expect a strong cancellation in light quark systems .
as shown in the original isgur - karl paper on the p - wave baryons @xcite , a very similar cancellation appears to occur at the two - body level in baryons .
however , unlike mesons , baryons can also experience three - body spin - orbit forces @xcite ( _ e.g. , _ potentials proportional to @xmath5 where @xmath6 , @xmath7 , and @xmath8 are the spin , position , and momentum of quark @xmath9 ) .
the matrix elements of these three body spin - orbit forces are all calculated in ref .
@xcite , but no apparent cancellation amongst them is found .
_ i.e. , _ the spin - orbit puzzle might more properly be called the baryon three - body spin - orbit puzzle " . in view of the facts that one could understand the smallness of spin - orbit forces in mesons and that the data also clearly called for small spin - orbit forces in baryons , the isgur - karl model anticipated a solution to the baryon three - body spin - orbit puzzle and _ as a first approximation _ discarded all spin - orbit forces .
it was assumed that , as in mesons , a more precise and broadly applicable description would have to treat residual spin - orbit interactions . in the meantime ,
a possible solution to this problem has been suggested @xcite in which relativistic effects enhance spin - spin over spin - orbit effects . this suggestion may prove to be correct , though it would then be an accident that in mesons a nonrelativistic solution presents itself . in this paper
i identify a special class of baryons which exhibit a meson - like solution to the spin - orbit puzzle and lead to a new perspective on this old problem .
consider a @xmath10 baryon in which the @xmath11 quark pair is compact and @xmath12 is far from their center - of - mass , as shown in fig .
as previously mentioned , the @xmath11 pair must be in a color @xmath1 , so the forces between it and @xmath12 are the same as those between an antiquark and @xmath12 .
if the internal dynamics of the @xmath11 pair were independent of @xmath13 , then each @xmath11 eigenstate would act as an extended quasi - antiquark with which @xmath12 could form a tower of meson - like excited states . the simplest such @xmath11 pair , and
the one which is the focus of this paper , has the @xmath11 pair in its isospin zero and spin zero ground state .
i label these state @xmath14 s since in them only the @xmath12 relative coordinate is excited over the ground state @xmath15 .
we will see that the only spin - dependent forces in these states are the spin - orbit forces experienced by @xmath12 , so they are a natural choice for a system in which to investigate the baryon spin - orbit puzzle .
= 2.8 in fig .
1 : ( a ) the relative coordinates @xmath16 and @xmath17 of a @xmath18 baryon with @xmath19 and @xmath20 .
( b ) a meson - like baryon configuration .
the idealization of a separation between @xmath21 and @xmath13 dynamics ( see fig .
1(a ) ) is actually realized in the zeroth - order states of the isgur - karl ( ik ) model @xcite where for @xmath22 and @xmath23 @xmath24 with @xmath25 and @xmath26 and @xmath27 spin - independent and spin - dependent potentials . in the ik model
one introduces an _ artificial _ harmonic term to define @xmath28 where @xmath29 and then treats @xmath30 and @xmath27 as perturbations on the zeroth - order hamiltonian @xmath31 which separates as advertized with @xmath32 and @xmath33 .
( the choice of @xmath34 and @xmath35 as variables was made historically so that in the su(3 ) limit @xmath36 the @xmath37 and @xmath38 oscillators become degenerate with wave functions that are manifestly good representations of the permutation group @xmath39 . )
note that by construction @xmath40 is an auxillary parameter which can be chosen to minimize the importance of @xmath30 . in what follows
i take @xmath41 where the first term is the color coulomb potential between quarks in a baryon ( to be contrasted with @xmath42 in a meson ) , @xmath43 is a constant , and @xmath44 is the long - range confining potential .
the zeroth - order eigenstates of ( [ eq : vharmonic ] ) are states with spatial wave functions @xmath45 and with the various flavor and spin states allowed by the generalized pauli principle ( given that the quarks are in the totally antisymmetric color state @xmath46 ) .
as stated earlier , i focus here on the isospin zero , light - quark - spin - zero states @xmath14 ( i.e. , those with flavor wave function @xmath47 and spin wave functions @xmath48 as defined in ref .
@xcite ) with the @xmath21 variable in its ground state : @xmath49 where @xmath50 is the normalized harmonic oscillator wave function for the variable @xmath51 with principal quantum number @xmath52 and angular momentum quantum numbers ( @xmath53 ) .
since @xmath13 is symmetric under @xmath54 interchange , these states all have the @xmath55 antisymmetry required by the generalized pauli principle , and the tower of @xmath13 excited states @xmath14 stand in one - to - one correspondence with the states of a @xmath56 meson with potential @xmath57 where @xmath58 and @xmath59 is a ficticious antiquark with spin and isospin zero and mass @xmath60 . in the harmonic limit
each of these towers of excited states have a spacing of @xmath61 where @xmath62 is the @xmath56 reduced mass . as mentioned earlier , @xmath40 is an auxillary parameter which may be chosen to minimize the perturbation @xmath30 . since , consistent with the @xmath55 symmetry of this system , eq . ( [ eq : hozero ] ) is trivially generalized to allow the @xmath63 spring constant to be distinct from the @xmath64 and @xmath65 spring constants , the @xmath37 and @xmath38 spring constants may be taken to be independent auxillary parameters @xmath66 and @xmath67 .
somewhat less trivial is the fact that @xmath67 may be chosen independently for each value of @xmath68 in the tower of @xmath38 excitations .
the subtowers consisting of states of fixed @xmath68 ( but excitation labelled by @xmath69 ) are all mutually orthogonal , so choosing @xmath67 to optimize the energy and wave function of @xmath70 is a good strategy for producing accurate orthonormalized eigenfunctions of the spin - independent hamiltonian .
( i optimize @xmath71 for each @xmath68 since these are the phenomenologically most relevant states ) . given this strategy , one never actually resorts to perturbation theory in @xmath30 for the @xmath71 states : their energies and eigenfunctions are best determined variationally .
the analogous strategy may obviously be employed for mesons . in an elementary @xmath56 system with
@xmath72 the expectation value of the spin - independent hamiltonian in @xmath73 is @xmath74 where @xmath75 .
table i shows how this spectrum varies with @xmath62 .
( throughout this paper i use the standard " parameters @xmath76 gev , @xmath77 gev , @xmath78 gev , @xmath79 gev , and @xmath80 gev@xmath81 , with @xmath82 since we are considering large - distance dominated systems . ) in a pure coulomb potential , the spacings between energy levels would grow like @xmath62 , while in a pure linear potential they would decrease like @xmath83 .
the latter behaviour can be seen at large @xmath84 , but at low @xmath84 the admixture of coulomb potential relevant to light quark spectroscopy leads to even more slowly varying splittings , as observed in nature ( see below ) . .
the variationally determined elementary meson spectrum ( in gev ) as a function of the reduced mass @xmath62 . [
cols="^,^,^,^,^,^",options="header " , ]
figure 1(b ) suggests that the tower of @xmath14 baryons i have described here will provide the simplest setting in which to begin to understand the baryon spin - orbit problem since , at least naively , they ought to have meson - like dynamics .
i have shown here that this expectation is in fact correct and that this select tower of baryons has no spin - orbit puzzle " .
the key to the solution of the puzzle for this select group of baryons is that for them the microscopic three - body spin - orbit forces described in the literature conspire to produce only meson - like quasi - two - body spin - orbit forces . as a result
these states experience the same strong cancellation between dynamical spin - orbit forces and thomas precession which leads to small spin - orbit effects in mesons .
it is possible that this work has no implications beyond identifying these special states .
however , in studying these states we have also developed a new perspective on the forces in baryons which may have a wider utility in confronting the baryon spin - orbit puzzle .
in particular , we have seen that the standard classification in the literature of spin - orbit forces as being two- and three - body is somewhat misleading . this classification is correct at the microscopic level ( see eq .
( [ eq:3bodyso ] ) ) , but not very relevant to the issue of whether the spin - orbit forces on a given quark @xmath12 appear to arise from the baryon center - of - mass , i.e. , whether they are quasi - two - body in character . with the spin - orbit puzzle solved for a slice through the baryon spectrum
, it would be odd if there were an intractable spin - orbit problem in baryons . at the least
, one may hope that the simple interpretation described here in terms of quasi - two - body ( i.e. , meson - like ) dynamics will lead to further insights into the spin - orbit puzzle .
in fact the arguments presented here for the @xmath14 baryons are immediately generalizable to the spin - orbit force on the third quark @xmath12 in @xmath85 and @xmath86 states , the total quark spin @xmath87 and @xmath88 @xmath89 baryons with the same spatial wave functions as @xmath90 .
since these three uds basis " states @xcite are sufficient to completely describe the su(3 ) baryon spectrum , it might seem that an even more substantial generalization is in hand : in the su(3 ) limit , baryon wave functions have permutation symmetry , so proving that quark 3 has meson - like spin - orbit forces is sufficient to prove that each quark does .
however , in the su(3 ) limit @xmath91 and our proofs are inapplicable : with such a degeneracy , leading order spin - orbit effects can occur _ via _ mixing between @xmath34 and @xmath35 excitations .
thus extending the considerations presented here beyond the spin - orbit force on @xmath12 in @xmath92 , @xmath85 , and @xmath86 will be nontrivial .
using our new perspective of quasi - two - body forces , we can visualize at least one obstacle to such extensions .
consider the counterparts to our @xmath14 state in which the @xmath34 variable is in a state with @xmath93 but @xmath94 .
in these circumstances the @xmath11 pair will develop an orbital color magnetic moment proportional to @xmath95 with which the color magnetic moment @xmath96 will in general interact .
it would thus seem to require a careful reanalysis of the effects of spin - orbit splittings on the baryon spectrum from this new perspective to determine if all of the _ observed _ nearly degenerate spin - orbit multiplets can be accommodated within it , and , if they can not be , to define the next steps along the path to solving this old problem .
it of course remains possible , despite the tantalizing fact that mesons and now the @xmath14 states admit a nonrelativistic solution to their spin - orbit splittings , that the ultimate solution is as suggested in ref.@xcite : that relativistic effects have simply produced a gross enhancement of spin - spin over spin - orbit interactions . for a general review of the isgur - karl model and for references , see n. isgur in _ modern topics in electron scattering _ , ed .
b. frois and i. sick ( world scientific , singapore , 1991 ) , p. 2 . for the discussion of spin - orbit forces see section v of n. isgur and g. karl , phys .
d*18 * , 4187 ( 1978 ) .
henriques , b.h .
kellet , and r.g .
moorhouse , phys .
64b * , 85 ( 1976 ) ; h. j. schnitzer , phys .
65b * , 239 ( 1976 ) ; * 69b * , 477 ( 1977 ) ; phys . rev . d*18 * , 3483 ( 1978 ) ; lai - him chan , phys . lett . *
71b * , 422 ( 1977 ) .
n. isgur and m.b .
wise , phys .
* b232 * , 113 ( 1989 ) ; phys . lett . * b237 * , 527 ( 1990 ) . for an overview of heavy quark symmetry see n. isgur and m.b .
wise , heavy quark symmetry " in _ b decays _
s. stone ( world scientific , singapore , 1991 ) , p. 158 , and in _ heavy flavors " _
buras and m. lindner ( world scientific , singapore , 1992 ) , p. 234 . for the original paper on heavy - light spectroscopy ,
see n. isgur and m.b .
wise , phys .
rev . lett . * 66 * , 1130 ( 1991 ) . | i describe a special class of meson - like @xmath0 excited states and present evidence supporting the similarity of their spin - independent spectra to those of mesons .
i then examine spin - dependent forces in these baryons , showing that predicted effects of spin - orbit forces are small for them for the same reason they are small for the analogous mesons : a fortuitous cancellation between large spin - orbit forces due to one - gluon - exchange and equally large inverted spin - orbit forces due to thomas precession in the confining potential .
in addition to eliminating the baryon spin - orbit puzzle in these states , this solution provides a new perspective on spin - orbit forces in all baryons .
epsf 1.0 cm |
by several measures ogle-2007-blg-224 was the most extreme microlensing event ( eme ) ever observed , having a substantially higher magnification , shorter - duration peak , and faster angular speed across the sky than any previous well - observed event .
these extreme features suggest an extreme lens .
fortunately , the observations themselves had extreme characteristics , and these permit one to test the nature of the lens .
ogle-2007-blg-224 ( ra=18:05:41 , dec=@xmath3:45:36 ) was announced as a probable microlensing event by the ogle collaboration @xcite on 9 may 2007 and independently by the moa collaboration @xcite on 12 may as moa-2007-blg-163 . at discovery
, it was already recognized to be quite short , einstein timescale @xmath4days , and was consistent with peaking at high magnification , although with very low probability of doing so .
such high - mag events ( typically defined as @xmath5 ) are extremely sensitive to planets , and hence planet - detection groups attempt to predict these rare events so as to intensively monitor them over their peak .
this is notoriously difficult , especially for short events , because so little information is available before peak .
indeed , four of the five high - mag events that have to date yielded planet detections had much longer - than - average einstein timescales , @xmath6days @xcite .
nevertheless , 20 hours before peak , ogle issued an alert calling this a `` possible high - magnification event '' , and 10 hours later , by combining ogle and moa data , including points from moa just 12 hours before peak , the @xmath7fun collaboration was able to predict @xmath8 and on this basis issued a general alert advocating intensive observations . as a result , on the night of the peak , the @xmath7fun bronberg 0.35 m telescope ( south africa ) obtained 754 unfiltered images over 6.5 hours ending at ut 03:43 just 10 minutes before peak , during which the magnified flux increased by a factor 38
. the @xmath7fun smarts 1.3 m telescope ( chile ) obtained a total of 304 images ( 52 in @xmath9 band , 8 in @xmath10 , and 244 in @xmath11 ) over 5.5 hours beginning 38 minutes before peak ; and the ogle 1.3 m warsaw telescope obtained 62 images ( 57 in @xmath9 and 5 in @xmath10 ) over 6.9 hours beginning 25 minutes before peak .
the robonet 2 m liverpool telescope obtained 8 @xmath12 images whose number and timing were determined by an automated program @xcite that queries several photometry databases and attempts to optimize observations .
critically , two of these observations were about 35 minutes before peak and two others were 25 minutes after peak .
additional observations were obtained by the moa 1.8 m telescope ( new zealand ) , the @xmath7fun 0.4 m auckland , 0.35 m farm cove , and 0.4 m vintage lane telescopes ( new zealand ) , the @xmath7fun 1.0 m mt .
lemon telescope ( arizona ) , the planet 1.0 m canopus ( tasmania ) , 0.6 m ( perth ) , 1.5 m boyden ( south africa ) and 1.54 m danish ( chile ) telescopes , as well as the robonet faulkes north 2 m ( hawaii ) , and faulkes south 2 m ( new south wales ) telescopes .
these observations helped in the determinations of the event s overall parameters , but not in the characterization of the peak , which is the central focus of this _
normally , microlensing events yield only one physical characteristic parameter , the einstein timescale @xmath13 , which is a degenerate combination of three physical parameters of the system , i.e. , the lens mass @xmath14 , the lens - source relative proper motion in the earth frame , @xmath15 , and the lens - source relative parallax @xmath16 , @xmath17 where @xmath18 and @xmath19 are the lens and source distances and @xmath20 . here
, @xmath21 is the `` microlens parallax '' , the ratio of the radius of the earth s orbit to the einstein radius projected on the observer plane , @xmath22 .
see figure 1 of @xcite for a geometric derivation of the relations between @xmath23 and @xmath24 . hence , if the lens is unseen , its mass and distance can be determined only by measuring both @xmath25 and @xmath26 @xcite . it is quite rare that either of these parameters can be measured , and only three times ( out of @xmath27 events ) have both been well - measured without seeing the lens itself @xcite .
however , as pointed out more than a decade ago , emes present a unique opportunity to measure both parameters @xcite . first , during an eme peak , the lens is very likely to transit the source , giving rise to distinctive `` finite - source '' effects that yield @xmath28 , the ratio of the source radius to the einstein radius . at high magnification , @xmath29 , where @xmath30 is the source - lens separation in units of @xmath25 .
since typically @xmath31 and @xmath32 , such effects are expected for @xmath33 . in the present case , we find that the lens transited the source with an impact parameter of @xmath34 , which permits a very accurate determination of @xmath35 .
we evaluate the source radius @xmath36 by extending the method of @xcite , to three bands , @xmath37 @xcite .
we obtain calibrated @xmath38 data from ogle - ii , and calibrated @xmath11 data by aligning smarts @xmath11-band to 2mass @xcite .
we measure the position of the clump and the source in 3 bands @xmath39 , @xmath40 , we adopt @xmath41 based on @xcite adjusted for the @xmath42-enhanced environment of the bulge @xcite .
we perform a monte carlo markov chain ( mcmc ) optimization of a model with 4 parameters : ( 1 ) a @xcite extinction law characterized by @xmath43 , ( 2 ) the mean visual extinction toward the clump @xmath44 , ( 3 ) the mean distance to the clump , @xmath45 , and ( 4 ) the extinction difference between the source and the clump @xmath46 .
we assume ( based on experience with high - resolution spectra of bulge dwarfs ) that @xmath47 , and we demand that the resulting @xmath48 be consistent with that predicted by @xcite from @xmath49 , with an error of 0.03 mag .
we then use the @xcite @xmath50 color / surface - brightness relation to derive @xmath51 , where the error also accounts for uncertainty in the model of the source flux ( which affects all three bands in tandem ) . as a check
, we note that this procedure yields @xmath52 , a typical value for the bulge , and @xmath53kpc , which is also consistent with most estimates ( keeping in mind that at @xmath54 , the bar is about 200 pc closer than the galactic center ) .
this result implies @xmath55 . combining this with the measured einstein timescale , @xmath56days , yields a proper motion in the earth frame @xcite of @xmath57 .
second , and more dramatically , emes are subject to `` terrestrial parallax '' effects @xcite . if a microlensing event is observed from two different locations , the source - lens relative trajectory will appear different : there will be a different impact parameter @xmath59 and a different time of closest approach @xmath60 .
since the projected einstein radius @xmath61 is typically of order au , the second observer should typically be in solar orbit to notice a significant effect , and indeed one such space - based measurement has been made @xcite .
however , the main way that microlens parallax has been measured in the past is to take advantage of the moving platform of the earth , but usually the event must last a large fraction of a year for this to work @xcite . for emes ,
the relevant spatial scale and timescales are reduced by a factor @xmath62 .
for example , if the projected velocity of the lens is @xmath63 in the westward direction , then @xmath60 will be about 80 seconds later in chile than south africa .
measuring the peak time of a normal microlensing event ( which typically lasts @xmath64days ) to this precision is virtually impossible .
but since the peaking timescale of this event is roughly the source crossing time , @xmath65minutes , such measurements become quite practical .
indeed the event passed both south africa and the canaries about 1 minute earlier than chile , with both time differences accurate to a few seconds .
see figure [ fig : lc ] . in practice
, we simultaneously account for the difference in impact parameters and peak times at all locations @xcite , as well as the earth s orbital motion ( which turns out be negligible ) , to measure both the microlens parallax @xmath66 and the direction of lens - source relative motion , @xmath67 south of west .
see figure [ fig : earth ] . the parallax measurement can also be expressed in terms of the offsets , @xmath68 , from which it is ultimately derived : @xmath69 @xmath70 @xmath71 where 1 = south africa , 2 = canaries , 3 = chile . the determinations of @xmath26 and @xmath25 yield the mass and relative parallax @xmath72 assuming ( as is consistent with its color and flux ) that the source is in the galactic bulge @xmath73 , the lens parallax and distance are @xmath74 , @xmath75 .
the projected velocity of the lens in the earth frame is @xmath76 . to find the projected velocity in the frame of the sun
, we must remove the motion of the earth around the sun ( @xmath77 , almost due east ) .
we then find @xmath78 at @xmath79 south of west .
this means that the lens is moving against the direction of galactic rotation , just @xmath80 out of the galactic plane ( toward galactic south ) .
taking account of the motion of the sun relative to the local standard of rest @xcite as well as the small mean motion ( and its uncertainty ) of the source , we find that the lens is moving at @xmath81 relative to the galactic disk at its location , almost directly counter to galactic rotation .
this motion is quite consistent with the kinematics of the galactic thick disk , which has an asymmetric drift of @xmath82 and a dispersion of @xmath83 in the direction of galactic rotation @xcite .
( it is also consistent with galactic halo stars , but with a probability more than 100 times smaller . )
moreover , since the inferred mass is below the threshold for burning hydrogen , the lens is almost certainly a brown dwarf .
while nearby brown dwarfs of this mass have been detected ( e.g. , @xcite ) , these are mostly quite young and so still retain the heat generated during their collapse from a cloud of gas .
the thick disk is of order 11 gyrs old , and hence brown dwarfs have had substantial time to cool .
only those very near the hydrogen burning limit are easy to see , and then only if they are relatively nearby .
_ hubble space telescope ( hst ) _ data were taken in @xmath10 , @xmath9 , @xmath84 , and @xmath11 bands @xmath85 days after peak when the source was magnified by a factor @xmath86 , and again almost exactly one year later , @xmath87 years . at the first epoch ,
the source and lens were virtually coincident , being separated by only @xmath88 , i.e. , @xmath89 pixels , while at the second epoch they were separated by @xmath90 pixel , where @xmath91 is the heliocentric proper motion .
these observations confirm that the excess flux ( above the source flux predicted from the microlensing fit ) is consistent with zero , but unfortunately not with very high precision because an unrelated star lying 150 mas from the source degrades the measurements .
we discuss the prospect for future observations of the lens flux below .
we now ask , given a standard galactic model @xcite , how likely it is that detailed information on a @xmath92day ( but otherwise unconstrained ) microlensing event toward this line of sight @xmath93 would turn out to imply a foreground ( @xmath94 ) thick - disk star rather than a star in the galactic bulge ?
the chance is just 1/235 .
hence , either we were extremely lucky to have found this object , or brown dwarfs are more common in the thick disk than our standard model for the mass function would predict .
specifically , our model assumes a relatively flat power law mass function for low - mass stars and brown dwarfs with @xmath95 from @xmath96 down to @xmath97 .
this model is consistent with results from observations of nearby thin disk brown dwarfs @xcite , and thus our detection would imply that brown dwarfs must be an order of magnitude more common in the thick disk than in the thin disk , in order that the a priori probability of detecting this event be @xmath98 .
this possibility should be considered seriously , since the prevalence of old brown dwarfs is essentially unconstrained by any data , owing to their faintness . in this light
, we note that two other sets of investigators have concluded that they must have been `` lucky '' unless old - population brown - dwarfs are more common than generally assumed .
first , @xcite discovered a halo brown dwarf , probably within 20 pc of the sun , a volume that contains only about 7 halo stars over the entire mass range from the hydrogen - burning limit to a solar mass @xcite . yet
the survey in which this object was discovered would only be sensitive to halo brown dwarfs in an extremely narrow mass range just below this limit @xcite .
second , @xcite analyzed a microlensing event of a nearby ( 1 kpc ) , bright ( @xmath99 ) source serendipitously discovered by an amateur astronomer hunting for comets , finding that it was most likely a low - mass star or brown dwarf moving at roughly @xmath63 .
based on the low rate of such events ( and the non - systematic character of the search ) , they concluded that they were `` probably lucky ... but perhaps not unreasonably so '' to have found this event .
see also @xcite .
finally , @xcite found 14 high - velocity objects in a sample of 332 m , l , and t dwarfs , which are consistent with thick - disk or halo kinematics .
only one of these has @xmath100 , which would be indicative of an old , low - mass brown dwarf @xcite , and with @xmath101 is most likely a halo brown dwarf , but could be in the high - velocity tail of the thick - disk . at absolute magnitude @xmath102 , it must be very close to the hydrogen - burning limit @xmath103 if it is as old as the thick disk and halo : @xmath104gyrs . as this object
was found within 10 pc , it is yet another `` lucky find '' ( compare to , e.g. , @xcite ) . while the evidence from each of these studies is marginal , collectively they point to the need for closer investigation of the frequency of old brown dwarfs .
brown dwarfs are very blue in @xmath105 and somewhat red in @xmath106 , so @xmath11 is the most favorable band to observe them . in the heliocentric frame ,
the lens is moving away from the source at @xmath107 .
hence , just 3 years after the event , the lens should be well removed from the source in keck telescope @xmath11-band images ( given its @xmath108 adaptive - optics point spread function ) .
equivalent angular resolution with _ hst _ would require waiting four times longer , although it would profit from the much fainter sky as seen from space as well as its better - defined point - spread function . from @xcite ( figs .
1,8,9,24 ) and @xcite ( fig .
23 ) one finds that the absolute magnitude of an 11 gyr old , @xmath109 brown dwarf is @xmath110 , implying @xmath111 at 525 pc ( assuming extinction @xmath112 ) . by comparison ,
the source is @xmath113 . in principle
, the lens could be younger , in which case it would be brighter and so easier to detect .
however , such high - velocity young stars are extremely rare , and there is no known mechanism for preferentially accelerating brown dwarfs to this speed . if the lens is indeed in the thick disk , this would be a challenging but not impossible observation , and would permit a unique test of models of old brown dwarfs by comparison with an object of known mass and well - constrained age .
the astrometry associated with this measurement would yield the lens - source relative proper motion @xmath114 , which would test both the _ magnitude _ of the microlens - based proper - motion and the _ direction _ of the microlens parallax , thereby confirming and refining ( or contradicting ) the microlens - based mass estimate @xcite .
looking further ahead , the space - based _ james webb space telescope _
( launch currently scheduled for 2013 ) will be able to take a spectrum of this object , thereby testing more detailed atmospheric models .
we thank c.s .
kochanek , s. poindexter and s. kozlowski for help with modeling the _ hst _ psf .
we acknowledge the following support : nsf ast-0757888 ( ag , sd ) ; nasa nng04gl51 g ( dd , ag , rp ) ; polish mnisw n20303032/4275 ( au ) ; hst - go-11311 ( ks ) ; nsf ast-0206189 , nasa naf5 - 13042 ( dpb ) ; korea science and engineering foundation grant 2009 - 008561 ( ch ) korea research foundation grant 2006 - 311-c00072 ( b - gp ) deutsche forschungsgemeinschaft ( csb ) ; pparc / stfc , eu fp6 programme `` angles '' ( w , njr ) ; pparc / stfc ( robonet ) ; dill faulkes educational trust ( faulkes telescope north ) ; grants jsps18253002 and jsps20340052 ( moa ) ; marsden fund of nz(iab , pcmy ) ; foundation for research science and technology of nz .
snodgrass , c. , tsapras , y. , street , r. , bramich , d. , horne , k. , dominik , m. , & allan , a. 2008 , in proceedings of the manchester microlensing conference : the 12th international conference and angles microlensing workshop , eds .
e. kerins , s. mao , n. rattenbury and l. wyrzykowski.pos(gmc8)056 | parallax is the most fundamental technique to measure distances to astronomical objects . although terrestrial parallax was pioneered over 2000 years ago by hipparchus ( ca
. 140 b.c.e . )
to measure the distance to the moon , the baseline of the earth is so small that terrestrial parallax can generally only be applied to objects in the solar system .
however , there exists a class of extreme gravitational microlensing events in which the effects of terrestrial parallax can be readily detected and so permit the measurement of the distance , mass , and transverse velocity of the lens . here
we report observations of the first such extreme microlensing event ogle-2007-blg-224 , from which we infer that the lens is a brown dwarf of mass @xmath0 , with a distance of @xmath1 and a transverse velocity of @xmath2 .
the velocity places the lens in the thick disk , making this the lowest - mass thick - disk brown dwarf detected so far .
follow - up observations may allow one to observe the light from the brown dwarf itself , thus serving as an important constraint for evolutionary models of these objects and potentially opening a new window on sub - stellar objects .
the low a priori probability of detecting a thick - disk brown dwarf in this event , when combined with additional evidence from other observations , suggests that old substellar objects may be more common than previously assumed . |
the maximal abelian gauge is obtained in @xmath0 gauge theory by maximizing the lattice functional @xmath1,\end{aligned}\ ] ] where @xmath2 and @xmath3 are the diagonal generators of the fundamental representation of @xmath0 .
the maximization is done by performing local @xmath0 gauge transformations on the links , which in practice is done using an @xmath4 subgroup update@xcite .
once the maximal abelian gauge is obtained , two independent abelian gauge fields are projected out , leaving @xmath5 gauge fields .
magnetic monopoles are then identified in the abelian configurations .
the lattice monopoles are used to directly compute the heavy quark potential@xcite .
the indirect center gauge was used in the first lattice studies of projected center vortices in @xmath4@xcite .
the indirect center gauge is obtained in the following way .
a non - abelian lattice gauge configuration is fixed to the maximal abelian gauge .
abelian projection is performed obtaining the abelian subgroup configurations : in @xmath4 this procedure obtains @xmath6 gauge fields , while in @xmath7 it yields @xmath8^{n-1}$ ] gauge fields .
the remaining @xmath6 configurations are then gauge fixed so that the gauge fields are brought as close as possible to the elements of the center group of the original @xmath7 theory . for the case of @xmath4
this corresponds to maximizing @xmath9 where @xmath10 parameterizes the abelian links .
the maximization is obtained by iteratively performing @xmath6 gauge transformations .
the final step to extract the vortices involves another projection of the lattice gauge fields onto the group elements of @xmath11 where @xmath12 is the unit matrix . in @xmath0 after maximal abelian gauge fixing and projection is done , the links are elements of @xmath13
. the @xmath13 links can be parameterized by the matrix : @xmath14 @xmath15 @xmath5 gauge transformations are then applied to these links to bring them as close as possible to the elements of the @xmath0 center group , @xmath16 the necessary gauge transformations can be parameterized using the fundamental , diagonal generators of @xmath0 : @xmath17 where some numerical factors are absorbed into the @xmath18 s and we use @xmath19 @xmath20 the center gauge in @xmath7 is obtained by maximizing the quantity @xmath21 for the indirect case in @xmath0 , this is accomplished by applying the @xmath5 gauge transformations described above to the diagonal @xmath13 links to maximize the quantity @xmath22 gauge transformations must be applied iteratively to each site independently to maximize the global quantity .
there are two local gauge conditions that must be met when the center gauge has been obtained .
the first condition comes from the @xmath23 contribution : @xmath24&=0,\\\end{aligned}\ ] ] where @xmath25 is the lattice derivative @xmath26 the second gauge condition to be met comes from the @xmath3 contribution : @xmath27&=0.\end{aligned}\ ] ] numerically , neither term can actually be zero . in practice ,
numerical iterations are continued until both terms are on the order of @xmath28 . in the continuum
these two gauge conditions are equivalent to the landau gauge : @xmath29 @xmath30
we generated 400 equilibrium @xmath0 lattice configurations on a @xmath31 lattice with @xmath32 .
at this value of @xmath33 the correlation length is about 4 lattice spacings . using these configurations and smearing techniques@xcite we computed the heavy quark potential and measured the full @xmath0 string tension , @xmath34 .
we found @xmath35 .
next , the 400 configurations were fixed to the maximal abelian gauge and the two species of abelian fields@xcite were projected out .
overrelaxation was used with @xmath36@xcite . in each abelian configuration dirac strings
were located using the @xmath37 plaquettes and the two species of monopole configurations were obtained .
the monopole configurations were used to compute the heavy quark potential ( see fig .
( 2 ) below ) .
results for both monopole species were found to be the same .
the string tension from the potential computed using monopoles was found to be @xmath38 .
this is about @xmath39 lower than the full @xmath0 string tension @xmath34 .
we believe this is the first quantitative calculation of the su(3 ) string tension from monopoles , although hints that the monopole string tension is low have appeared before@xcite .
after gathering the abelian configurations , we next applied our abelian center gauge fixing algorithm .
no closed form solution of the parameters @xmath40 and @xmath41 could be found for the transcendental equations resulting from maximizing @xmath42 .
the @xmath18 s were determined iteratively .
the value of the normalized abelian trace computed from using the @xmath13 links was observed to change from an average value of 0.525 before center gauge fixing to 0.922 after gauge fixing . from the center gauge fixed configurations ,
the links were replaced by the @xmath43 phase each was closest to .
wilson loops were computed using the @xmath43 configurations .
the wilson loop data is presented in figure ( 1 ) .
fits to the potential computed using the @xmath43 wilson loops gave a string tension of @xmath44 .
thus , we find the string tension from indirect center projection to be too low . in figure ( 2 )
we compare the heavy quark potentials .
the potentials computed from the monopoles and projected center vortices show little coulomb behavior at small @xmath45 as compared to the full @xmath0 potential .
this has been observed in other calculations in @xmath4 and @xmath6@xcite .
calculations which demonstrate the confinement mechanism are not required to reproduce the short distance behavior of the potential , only the string tension .
we have presented results for the string tension computed using monopoles and center vortices .
the calculations use operations defined at the level of a single lattice spacing .
neither monopoles nor center vortices reproduce the full @xmath0 string tension .
there is still a need to refine the methods used here if one is to discard or accept monopoles or center vortices as the confinement mechanism in qcd .
we conclude by discussing possible reasons to suspect systematic errors exist in these calculations .
monopoles and vortices are computed using objects defined on the single link level .
the results for the string tension may be corrupted by unphysical degrees of freedom due to misidentification of monopoles and center vortices by using operators defined at the single link level .
also , we may not be using volumes large enough to suppress finite volume effects since monopoles and center vortices are objects that likely extend over many lattice spacings . to investigate the first problem
, we need to construct other methods for identifying monopoles and center vortices . for the second problem
we need to work on larger lattices .
using larger lattices is difficult due to the computationally intensive nature of the gauge fixing methods .
the numerical procedures for the maximal abelian and center gauge fixing suffer from gribov copy problems .
in particular , for the direct and indirect center gauge fixing procedures in @xmath4 the effects of gribov copies on the numerical value of the string tension are large@xcite .
the effect of taking account of the gribov ambiguity for @xmath4 for all cases investigated so far , is to decrease the corresponding string tension .
so for monopoles in the maximal abelian gauge , the monopole string tension for one gauge fix / configuration is quite close to the full @xmath4 value @xcite , and decreases by @xmath46 when account is taken of the gribov ambiguity by generating many copies / configuration @xcite .
our result for the @xmath0 monopole string tension is considerably lower than the full @xmath0 value even at one gauge fix / configuration , and would presumably be reduced somewhat more when gauge ambiguities are accounted for .
thus the @xmath4 and @xmath0 cases appear to be rather different . given the large noise level of our data , we did not succeed in extracting a string tension at the @xmath5 level , before projecting to monopoles or vortices .
given our low results for the monopole and center vortex string tensions , this is now an important quantity to measure . a failure of abelian projection at the @xmath13 level for @xmath0 would have important implications for ideas on confinement in @xmath0 .
we plan to return to this question in future work .
9 n. cabibbo and e. marinari , phys .
b119 ( 1982 ) 387 .
t. banks , r. myerson and j. kogut , nucl .
b 129 ( 1977 ) 493 .
john d. stack , steven d. neiman and roy j. wensley , phys .
d50 ( 1994 ) 3399 .
yoshimi matsubara , sawut ilyar , tsuyoshi okude , kenji yotsuji , tsuneo suzuki , nucl .
42 ( 1995 ) 529 - 531 . l. del debbio , m. faber , j. greensite , and s. olejnik , phys .
d55 ( 1997 ) 2298 .
the ape collaboration(m .
albanese et .
al ) , phys .
lett.*192b * ( 1 987 ) 163 .
k. yee , mod .
. lett . *
a * ( 1994 ) 1991 .
j. mandula and m. ogilvie , phys .
b * 248 * ( 1990 ) 156 .
t. okude , et .
al , nucl . phys .
* b533 * ( 1998 ) 576 j.d . stack and r.j .
wensley , nucl . phys .
* b371 * ( 1992 ) 597 .
bornyakov , d.a .
komarov , m.i .
polikarpov , a.i .
veselov , jetp lett .
71 ( 2000 ) 231 - 234 .
g. s. bali , v. bornyakov , m. muller - preussker , and k. schilling , phys .
d54 ( 1996 ) 2863 . | we present results for the heavy quark potential computed in su(3 ) from magnetic monopoles and from center vortices .
the monopoles are identified after fixing su(3 ) lattice configurations to the maximal abelian gauge .
the center vortices are identified after using an indirect center gauge fixing scheme which we describe for su(3 ) .
z(3 ) center vortices are extracted and used to compute the potential .
the values of the string tensions from monopoles and vortices are compared to the full su(3 ) string tension .
epsf a clear understanding of the confinement mechanism in qcd is still elusive .
two particular mechanisms have gained the most attention : magnetic monopoles and center vortices . here
we present results of calculations in su(3 ) pure gauge theory of the heavy quark potential using methods which test these two different mechanisms . |
let @xmath0 be a riemannian manifold and @xmath1 a diffeomorphism . an @xmath2-invariant set @xmath3 has a dominated splitting if if its tangent bundle @xmath4 , where @xmath5 and @xmath6 are non trivial invariant continuous subbundles such that , for constants @xmath7 and @xmath8 : @xmath9 clearly , any hyperbolic splitting is a dominated one
. some important consequences of this property on the dynamics were given by pujals and sambarino in @xcite .
a spectral decomposition theorem was obtained for @xmath10 compact surface diffeomorphisms having dominated splitting over the limit set @xmath11 where @xmath12 and @xmath13 are the @xmath14 and @xmath15-limit sets of @xmath16 , respectively : * theorem ( pujals - sambarino @xcite ) : * _ let @xmath0 be a compact 2-manifold and @xmath17 a @xmath18-diffeomorphism .
assume that @xmath19 has a dominated splitting .
then @xmath19 can be decomposed into @xmath20 such that : _ 1 .
@xmath21 is a set of periodic points with bounded periods contained in a disjoint union of finitely many normally hyperbolic periodic arcs or simple closed curves .
@xmath22 is a finite union of normally hyperbolic periodic simple closed curves supporting an irrational rotation .
_ @xmath23 can be decomposed into a disjoint union of finitely many compact invariant and transitive sets ( called basic sets ) .
the periodic points are dense in @xmath23 and at most finitely many of them are non - hyperbolic periodic points .
the ( basic ) sets above are the union of finitely many ( nontrivial ) homoclinic classes .
furthermore @xmath24 is expansive . _
our purpose , in this work , is to construct simple examples of dynamical systems with attractors admitting a dominated splitting and where the dynamics is of type @xmath21 or @xmath22 .
we will follow the ideas developed in @xcite and @xcite . in @xcite ,
non conservative billiards were introduced by a modification of the reflection rule and the existence of attractors was demonstrated for a wide class of dispersing and semi - dispersing billiards and of billiards with focusing components . in @xcite models
were studied numerically and different attractors , periodic and chaotic are presented .
we concentrate on strictly convex billiard tables , with boundary formed by a unique sufficiently differentiable focusing component .
they present structures like kam stability islands and invariant rotational curves non homotopic to a point .
we will investigate how , in the presence of non conservative perturbations , an invariant curve will give rise to an attractor .
the maps we consider here are more general than the pinball billiards introduced in @xcite , as the perturbation of the angle is not necessarily biased to the normal direction .
the contents of the present paper is outlined in four sections that follow this introduction . in section [ sec : cones ] we present the main tools needed to work with dominated splitting while section [ sec : classic ] deals with the basic properties of classical billiards on ovals . in section [ sec : nonela ] we introduce our _ non elastic billiards_. they are defined as a composition of a classical billiard followed by a change of the reflection angle , corresponding to a contraction in the vertical fibers of an invariant rotational curve .
we will prove that under some differentiability hypotheses and some bounds on the contraction , there exists a compact strip in the phase space , such that the non elastic billiard map is a @xmath25diffeomorphism from that strip onto its image .
its limit set contains the invariant curve and has a dominated splitting .
moreover , the non elastic dynamics on the invariant curve is determined by its rotation number with respect to the original classical billiard map .
this result will guide us to construct our examples of non elastic billiards on ovals with dominated splitting and attractors supporting a rational or an irrational rotation .
they are presented in section [ sec : exa ] , where we explore their properties theoretically and numerically .
our result ( theorem [ nonelastic ] ) could clearly be established in the more general setting of @xmath18 conservative twist maps .
once we have a @xmath18 invariant rotational curve and assuming the necessary bounds , everything would work likewise in the proof . the main problem is to build specific examples , other than the obvious twist integrable case ( which corresponds to the circular billiard ) or the oval billiards presented here , and check , for instance and at least numerically , the size of the basin of attraction .
the existence of a dominated splitting follows from the existence of an eventually strictly invariant cone field and of an uniform control of expansions and/or contractions as showed in @xcite .
the criteria presented there in proposition 4.1 can be translated , in dimension 2 , into the operational lemma below .
let @xmath26 be two vector fields such that @xmath27 and @xmath28 are linearly independent vectors in @xmath29 .
they induce a nondegenerate quadratic form @xmath30 on @xmath31 by @xmath32 and a cone field given at each @xmath16 by @xmath33 and whose boundaries , at each point , are given by @xmath34 .
if the vector fields are continuous , the quadratic form and the cone field are also continuous . given @xmath35 and a vector @xmath36 , let @xmath37 denote the image of @xmath38 under the derivative @xmath39 .
then we have @xmath40_u \left ( \begin{array}{c } a \\ b \end{array } \right ) \label{eq : ab}\ ] ] where @xmath41_u$ ] is the matrix representation of the derivative at @xmath16 , with the choice of @xmath42 and @xmath43 as bases of @xmath44 and @xmath45 respectively . [ lema1 ]
let @xmath3 be a compact @xmath2-invariant subset of @xmath0 . if there is a choice of vector fields @xmath46 such that the entries of @xmath41_u$ ] are strictly positive for every @xmath47 , then @xmath3 has a dominated splitting . proof : if the entries of @xmath41_u$ ] are strictly positive for every @xmath47 then for every @xmath48 , @xmath47 , @xmath49 , @xmath50 we have @xmath51 where @xmath52 .
this implies that for every @xmath47 @xmath53 , i.e. , @xmath54 is strictly @xmath55-invariant ( @xmath2 is strictly @xmath30-separated ) .
it follows from proposition 4.1 in @xcite that @xmath3 has a dominated splitting .
let @xmath57 be an oval , i.e. , a plane , simple , closed , @xmath58 , curve , with strictly positive curvature , parameterized counterclockwise by @xmath59 , the angle between the tangent vector and an horizontal axis .
let @xmath60 be its radius of curvature at @xmath61 .
the classical billiard problem on @xmath57 consists of the free motion of a point particle inside @xmath57 , making elastic reflections at the impacts with the boundary .
the motion is then determined by the point of reflection at @xmath57 and the direction of motion immediately after each reflection .
they can be given by the parameter @xmath62 , that will locate the point of reflection and by the angle @xmath63 between the tangent vector and the outgoing trajectory , measured counterclockwise .
the classical billiard defines a map @xmath64 from the open cylinder @xmath65 into itself , which has some very well known properties ( see , for instance , @xcite and @xcite for the properties of billiards and twist maps listed in this section ) . denoting @xmath66 ,
the derivative of @xmath64 at @xmath67 is @xmath68 where @xmath69 and @xmath70 is the distance between @xmath71 and @xmath72 .
if @xmath73 is @xmath74 , then @xmath64 is a @xmath75-diffeomorphism .
it preserves the measure @xmath76 .
it is reversible with respect to the reversing symmetry @xmath77 , as @xmath78 , meaning that the phase space is symmetric with respect to the line @xmath79 .
moreover , as @xmath57 is an oval , @xmath64 has the monotone twist property with rotation interval @xmath80
. for each @xmath81 there exists a closed , invariant , minimal set @xmath82 , which can be injectively projected on @xmath83 and such that the induced dynamics preserves the order of @xmath84 .
this invariant set @xmath85 can be a periodic orbit , an aubry mather set or a rotational curve , i.e. , a continuous closed curve which is the graph of a lipschitz function .
we will concentrate on billiards having an invariant rotational curve @xmath86 graph@xmath87 , @xmath88 a lipschitz function .
in general , for a given boundary @xmath57 , the set of rotation numbers @xmath89 such that @xmath85 is an invariant rotational curve is nowhere dense in @xmath80 , but nevertheless invariant rotational curves exist on a large class of classical oval billiards . for instance
, the circular and the elliptical billiards have invariant rotational curves with any rotation number @xmath89 .
the billiard on a curve of constant width have the horizontal curve @xmath90 as an invariant rotational curve with rotation number @xmath91 .
but , apart from special examples like these , it is difficult to find billiards with invariant rotational curves of rational rotation number .
in fact , having an invariant rotational curve with rational rotation number is not a generic property for billiards on ovals .
the generic dynamics is to have , for each rational rotation number , a finite number of periodic orbits with this rotation number and at least one of them is hyperbolic , with transverse homoclinic orbits @xcite . from the other side ,
having an invariant rotational curve with irrational rotation number is quite general .
if the oval boundary @xmath73 is @xmath92 then lazutkin theorem guarantees a whole family of invariant rotational curves with diophantine @xmath89 near the boundaries of the cylinder @xmath93 ( @xcite , @xcite ) . and sufficiently small perturbations of the circular billiard present , in addition to the diophantine invariant rotational curves , given by the kam theorem , uncountably many invariant rotational curves with liouvillian rotation number . for our purposes
, we will need the curve @xmath94 , or equivalentely the function @xmath95 , to be @xmath18 or more .
this is the case of the circular , the elliptical and the constant width billiards , for instance , where the rotational invariant curves are , in fact , analytic . also , the curves guaranteed by lazutkin theorem are as differentiable as the billiard map , so @xmath96 or more .
but , as showed in @xcite , invariant rotational curves can be just a little bit more regular than lipschitz , and so we have to impose the existence of a @xmath18 rotational invariant curve as a condition to the billiards we will use . in the next paragraphs we present some bounds that will be necessary to prove our results .
let us assume that the invariant set @xmath85 is a closed curve @xmath94= graph@xmath87 . as @xmath97 preserves the order of @xmath98 then either @xmath90 or there exist constants @xmath99 and @xmath64 such that @xmath100 or @xmath101 .
the @xmath18 ( or more ) character of @xmath94 also implies that a tangent vector @xmath102 is sent by @xmath103 on a tangent vector to @xmath94 at @xmath104 .
the preservation of orientation implies that the first coordinate of @xmath105 must be strictly positive .
so @xmath106-r_0\sin\alpha_0>0 \ \ \hbox { and } \ \ 1+g'(\varphi_0)>\frac{r_0\sin\alpha_0}{l}>0\ ] ] this implies that @xmath107 for every @xmath61 . as the billiard is reversible , the graph of @xmath108 is also a rotational invariant curve and then @xmath109 for every @xmath61 .
so , for any @xmath18 invariant @xmath86graph@xmath87 , @xmath110 . to each invariant rotational curve @xmath94 is associated a caustic @xcite , a curve lying inside the billiard table and tangent to every segment of the billiard trajectory between two consecutive impacts .
if @xmath67 and so @xmath111 belong to @xmath94 , then @xmath112 , @xmath113 and we have @xmath114 where the two terms on the left hand side are strictly positive .
the first one measures the distance to the tangency point ( on the caustic ) from the initial point @xmath71 and the second one , from the final point @xmath72 ( see @xcite ) .
let @xmath115 be a @xmath18 classical billiard map on an oval , with a @xmath18 invariant rotational curve @xmath116 , given by the graph of @xmath117 .
a compact subset of the phase space @xmath118 with non - empty interior and whose boundaries are two distinct rotational curves ( not necessarily invariant nor graphs ) will be called a compact strip .
let @xmath119 be a closed interval containing @xmath120 .
given @xmath121 , a @xmath18 strictly increasing contraction with @xmath122 , we can define a non elastic billiard map @xmath123 on a compact strip @xmath124 containing @xmath116 by @xmath125 with @xmath124 chosen such that if @xmath126 then @xmath127 .
@xmath123 is the composition of a classical billiard followed by a change at the reflection angle , corresponding to a contraction in the vertical fibers of the invariant rotational curve @xmath116 .
observe that @xmath128 corresponds to the classical unperturbed billiard and that @xmath129 corresponds to a map that sends all the points in @xmath124 onto the invariant curve ( called _ slap billiard _ in @xcite ) .
the derivative of @xmath123 is given by @xmath130 where @xmath131 and @xmath132 our main result is : [ nonelastic ] given a classical oval billiard map @xmath64 , with a @xmath18 invariant rotational curve @xmath133 , consider a compact strip @xmath124 containing @xmath116 and a closed interval @xmath134 , such that @xmath127 if @xmath126 . if @xmath135 is a @xmath18 function satisfying @xmath122 and @xmath136 ( with @xmath137 depending only on @xmath116 ) , then there exists a compact strip @xmath138 such that the non elastic billiard map @xmath123 defined by @xmath64 , @xmath95 and @xmath139 is a @xmath18-diffeomorphism from @xmath140 onto @xmath141 .
its limit set @xmath142 contains @xmath116 and has a dominated splitting .
moreover , the non elastic perturbation does not change the dynamics on @xmath116 .
proof : the non elastic billiard @xmath143 is the composition of the @xmath18 classical billiard map @xmath64 with the @xmath18 perturbation of the identity @xmath144 , where @xmath139 is a @xmath18 contraction . then @xmath123 is a @xmath18 diffeomorphism . given @xmath145 , let @xmath146 and @xmath147 be two linearly independent vector fields defining the cone field @xmath148 and the associated quadratic form @xmath149 ( as in section [ sec : cones ] ) . using the change of bases matrices
, we have @xmath150_u = { \displaystyle \frac{1}{2\delta r_1 \sin \alpha_1 } } \left ( \begin{array}{cc } \delta ( l_{0 } - \delta l ) + ( 1- h'_1 ) ( \delta l_{1 } - l_{01 } ) & \delta ( l_{0 } + \delta l ) -(1- h'_1 ) ( \delta l_{1 } + l_{01 } ) \\ \delta ( l_{0 } - \delta l ) - ( 1- h'_1 ) ( \delta l_{1 } - l_{01 } ) & \delta ( l_{0 } + \delta l ) + ( 1- h'_1 ) ( \delta l_{1 } + l_{01 } ) \end{array } \right ) \label{eqn : dp}\ ] ] where @xmath151 relation ( [ eq : caustica ] ) implies that for every @xmath67 and @xmath104 on @xmath116 we have @xmath152 the billiard boundary @xmath57 is an oval and as it is compact there exist constants @xmath153 and @xmath154 and a width @xmath155 such that @xmath156 and @xmath157 . as the invariant curve is also compact , for every @xmath67 and @xmath104 on @xmath116 , there exist constants @xmath158 and @xmath159 such that @xmath160 and @xmath161 .
so for points on @xmath116 there are constants @xmath162 and @xmath163 such that @xmath164 by formula ( [ eqn : dp ] ) , each entry of the matrix @xmath165 _ u$ ] for @xmath166 is of the form @xmath167 now , if @xmath136 , we can choose @xmath168 such that @xmath169 . as @xmath123 is a @xmath18 diffeomorphism and remembering that @xmath116 is compact , we can find a strip @xmath170 , containing @xmath116 , where @xmath123 is well defined and all the entries of @xmath165 _ u$ ] are strictly positive .
then , by lemma [ lema1 ] , @xmath171 has a dominated splitting and contains @xmath116 , since @xmath122 .
moreover , @xmath172 and the dynamics under @xmath123 , on @xmath116 , is the same as under @xmath64 .
as @xmath116 and @xmath123 are @xmath18 , for our billiard dynamics , @xmath116 is either a set of periodic points of same period linked by homo / heteroclinic arcs or supports a rational or an irrational rotation .
.4 cm @xmath56 this result will guide us to construct examples of non elastic billiards on ovals with limit set having a dominated splitting and supporting a rational or an irrational rotation ( pieces of type @xmath173 or @xmath174 of pujals - sambarino s theorem ) .
this will be done in the next section .
we will pay attention to the maximal possible size of the strip @xmath140 and will try to see if there are other attractors on @xmath142 than @xmath116 .
the simplest example of a classical billiard with invariant rotational curves is the circular one .
this billiard map is linear and is given by @xmath175 .
the phase space @xmath65 is foliated by invariant horizontal curves and the dynamics on each one of them is simply a rotation of @xmath176 .
fix @xmath183 , a closed interval with @xmath184 and @xmath185 , any @xmath18 strictly increasing contraction such that @xmath122 and @xmath186 .
the non elastic billiard @xmath123 is defined on the strip @xmath187 and is given by @xmath188 .
now , we choose @xmath191 and @xmath192 such that @xmath193 $ ] is the biggest horizontal straight strip contained in @xmath140 . as each boundary @xmath194 is invariant under @xmath64 ,
we have that @xmath195 .
the map @xmath123 is a horizontal rotation followed by a vertical contraction . denoting @xmath196 , it is easy to see that @xmath197 , as @xmath198 and so the horizontal circle @xmath116 is the unique attractor of @xmath123 in @xmath199
. moreover , the restricted map @xmath200 is just a rotation of angle @xmath201 . if @xmath202 is rational , @xmath116 is a normally hyperbolic simple closed curve composed by periodic points of the same period .
if @xmath202 is irrational @xmath116 is a normally hyperbolic closed curve supporting an irrational rotation .
this yields an example of a diffeomorphism , defined on a strip @xmath199 , whose limit set has dominated splitting and is composed by a unique piece of type @xmath173 or @xmath174 of pujals - sambarino s theorem . clearly , the size of the strip @xmath199 depends on the choice of the contraction @xmath139 .
taking for instance @xmath203 , with @xmath204 , the non elastic billiard is given by @xmath205 and the basin of attraction of @xmath116 contains any straight strip @xmath193 $ ] .
a similar example is given by the classical elliptical billiard .
we consider an ellipse @xmath57 with eccentricity @xmath206 and minor axis 1 .
its radius of curvature @xmath207 satisfies @xmath208 .
the associated classical billiard map is denoted by @xmath209 .
this billiard system is integrable : the function @xmath210 is a first integral ( see , for instance @xcite ) and @xmath118 is foliated by the levels of @xmath211 , with @xmath212 .
if @xmath213 , the level set consists of two invariant , analytic and symmetric rotational curves , the lower one contained in @xmath214 and the upper one in @xmath215 .
but , unlike the circular case , @xmath64 has two elliptic islands of period 2 , obstructing the rotational invariant curves to foliate the whole phase - space , as can be seen on figure 1(left ) .
it also has a hyperbolic 2-periodic orbit , with a saddle connection , corresponding to the level @xmath216 .
we have then that , for any @xmath220 , @xmath221 .
differentiating twice with respect to @xmath61 we get @xmath222 and @xmath223 .
the extremal points of @xmath224 must satisfy @xmath225 which implies @xmath226 .
we can then take @xmath227 the associated non elastic billiard map is given by @xmath228 where the contraction @xmath229 is an arbitrary @xmath18 function satisfying @xmath122 and @xmath230 , and @xmath183 is a closed interval containing @xmath120 .
then , by theorem [ nonelastic ] , there exists a compact strip @xmath140 , containing @xmath116 , such that @xmath189 is a @xmath18 diffeomorphism , @xmath190 has a dominated splitting and contains @xmath116 .
let @xmath231 be two constants of motion satisfying @xmath232 , and @xmath233=graph@xmath234 be the lowest invariant rotational curves associated to @xmath231 , respectively .
we also suppose that @xmath235 is the biggest strip of this type contained on @xmath140 .
as @xmath233 are invariant under @xmath64 we have @xmath195 . for @xmath236
, we denote @xmath196 .
let us suppose , for instance , that @xmath237 , the other case being analogous .
as @xmath123 is a translation on a @xmath64-invariant rotational curve followed by a contraction on the vertical direction toward @xmath116 , then @xmath238 and @xmath116 is the @xmath14-limit of any @xmath236 .
as in the circular case , @xmath116 is the unique attractor of @xmath239 . using action - angle variables ,
chang and friedberg @xcite have shown how to decide if the rotation number associated to the level @xmath240 of a given initial condition @xmath67 , is rational or irrational and then determine the dynamics on each level curve .
as @xmath241 , this allows us to choose @xmath242 in order to have , as the unique attractor , a normally hyperbolic closed curve supporting an irrational rotation or a normally hyperbolic closed curve composed by periodic points of same period .
although theoretically promising as a result , depending on the choice of @xmath116 and @xmath139 , the strip @xmath199 can be very thin .
in particular it will never contain points of the elliptical islands . in trying to go beyond the theoretical predictions and find examples of non elastic elliptical billiards defined on a bigger part of the phase space
, we shall remember that weak contractions will never destroy the rotation carried by the linear ellipticity of the 2-periodic orbit .
so we can not expect the associated non elastic billiard to be defined on a strip containing the islands .
however , it is an interesting question whether this can be achieved by taking a sufficiently strong contraction .
we present some numerical simulations where this can be done .
we choose an ellipse with @xmath243 and fix the invariant curve @xmath116 at @xmath244 .
our choices of the eccentricity and @xmath116 are rather arbitrary .
figure [ fig : elipse1 ] ( left ) below diplays the phase space of the classical elliptical billiard . the invariant curve @xmath116 is enhanced .
the horizontal axis corresponds to @xmath245 and the vertical to @xmath63 , the left bottom corner being the origin .
we consider linear perturbations @xmath246 , with @xmath204 .
@xmath247 implies that there is no perturbation ( classical billiard ) and @xmath248 implies that all the points in the phase space land on the invariant curve after one iteration ( slap billiard ) .
the non elastic billiard @xmath249 is a @xmath18 diffeomorphism on any compact strip contained on @xmath65 . calculating @xmath137 by formula ( [ underline ] )
we get that if @xmath250 there is a strip @xmath199 such that @xmath116 is the unique attractor of @xmath123 on @xmath199 and has a dominated splitting .
figure [ fig : elipse1 ] ( right ) illustrates the basin of attraction of the curve @xmath116 for @xmath251 : black points correspond to initial conditions which approach @xmath116 under iteration .
this simulation thus indicates that the basin of attraction of @xmath116 is the whole phase space , ie , that @xmath199 can be any compact strip contained in @xmath65 and @xmath116 is the unique attractor of @xmath123 .
note that , as we have mentioned before , either @xmath116 is composed of periodic points of same period or supports an irrational rotation .
there are no other known @xmath18 ovals such that the classical billiard map is integrable , other than the circle and the ellipse .
but it is well known that there are oval billiards with @xmath18 invariant rotational curves .
taking a sufficiently differentiable , non integrable , classical oval billiard map @xmath64 with a @xmath18 invariant rotational curve @xmath116=graph@xmath87 we can , as before , take a contraction @xmath139 satisfying the hypothesis of theorem [ nonelastic ] and define a non elastic billiard map @xmath123 . then there exists a strip @xmath140 on which @xmath123 is a @xmath18 diffeomorphism on @xmath140 , @xmath142 contains @xmath116 and has a dominated splitting . depending on the rotation number of
@xmath116 we can have , as in the previous examples , a normally hyperbolic closed curve supporting an irrational rotation or a normally hyperbolic closed curve composed by periodic points of same period .
but we can no longer guarantee that the only attractor of @xmath123 is @xmath116 . in order to explore numerically what happens in those more general cases
, we have to pick a concrete example .
although in any example one can prove the existence of whole families of invariant curves , it is almost impossible for any fixed curve , to write down the function @xmath95 for which it is the graph . in general
this can only be achieved in very specific examples as , for instance , the constant width curves , where @xmath252 .
we will deal with a much richer example , given by symmetric perturbations of the circle .
let @xmath253 be the oval parameterized by the angle @xmath61 and which radius of curvature is of the form @xmath254 , @xmath255 and @xmath256 .
tabachnikov ( section 2.11 , @xcite ) showed that its associated classical billiard map have an invariant straight line given by @xmath257 if @xmath258 satisfies @xmath259 .
it is not difficult to show that the dynamics on @xmath116 is an irrational rotation @xcite .
for @xmath260 , @xmath261 satisfies this condition and the line @xmath116 given by @xmath262 is invariant .
figure [ fig : mesa ] displays the billiard table and the corresponding phase space of @xmath263 defined by @xmath264 . as in the previous section , the horizontal axis corresponds to @xmath245 and the vertical to @xmath63 , the left bottom corner being the origin .
note the invariant straight line @xmath116 . as in the other numerical examples , we consider linear perturbations @xmath246 , @xmath204 .
on @xmath116 , we have @xmath265 and @xmath266 and we can take @xmath267 . then , if @xmath268 , there exists a strip @xmath140 on which @xmath123 is a @xmath18 diffeomorphism , @xmath142 contains @xmath116 and has a dominated splitting .
figure [ fig:4bacias ] illustrates the basin of attraction of @xmath116 for @xmath269 and @xmath270 : black points correspond to initial conditions which approach @xmath116 under iteration .
the white tadpoles correspond to points attracted to the 6-periodic orbits , linearly elliptic for the original classical billiard .
we observe that as the contraction factor @xmath271 is increased , the basin of attraction of @xmath116 grows until it eventually occupies the whole phase space .
thus , for strong contractions , we may have @xmath123 defined on any compact strip @xmath140 and having @xmath116 as its unique attractor .
any strictly convex classical billiard map is a monotone twist map with rotation interval @xmath80 . to each @xmath273
is associated a set @xmath85 .
if @xmath89 is irrational , @xmath274 is either a rotational invariant curve or an aubry - mather set .
an aubry mather set is a closed , invariant , minimal set , projecting injectively on a cantor set of @xmath275 and such that the dynamics preserves the order of @xmath276 .
it is contained in a non invariant graph of a continuous piecewise linear lipschitz function @xmath117 ( see , for instance @xcite , section 13.2 ) .
let us take a classical billiard map @xmath64 with two rotational invariant curves @xmath277 and @xmath278 and an aubry - mather set @xmath279 contained in the strip bounded by @xmath277 and @xmath278 and a perturbation @xmath139 on this strip .
the non elastic billiard map can be then defined as before , as @xmath280 , where @xmath281 .
as @xmath139 is a contraction , the limit set of @xmath123 will contain the aubry - mather set @xmath279 .
however , we must remark that @xmath123 is not @xmath18 because @xmath95 is not even a differentiable function .
then , we can not apply pujal - sambarino s theorem .
99 arnaud m c 2009 _ publ .
hautes etudes sci _ * 109 * 117 arroyo a , markarian r and sanders d p 2009 _ nonlinearity _ * 22 * 14991522 berry m 1981 _ eur
. j. phys . _
* 2 * 91102 birkhoff g d 1932 _ bull .
smf _ * 60 * 126 chang s j and friedberg r 1988 _ j. math . phys . _
* 29 * 15371550 chernov
n i and markarian r 2006 _ chaotic billiards .
_ mathematical surveys and monographs * 127 * .
american mathematical society .
dias carneiro m j , kamphorst s o and pinto - de - carvalho s 2007 _ nonlinearity _ * 20 * 24532462 r.douady : applications du thorme des tores invariants . thse de 3me cycle , univ .
paris vii , 1982 .
ferreira , g. : billiards maps with invariant straight lines and curves .
book of abstracts of the second international conference billiards2011 , 2011 , http://www.rc.unesp.br/igce/demac/billiards11/ katok a and hasselblatt b 1997 _ introduction to the modern theory of dynamical systems_. ( cambridge univ . press ) katok , a. , hasselblatt , b. : _ principal structures _
, in handbook of dynamical systems , vol 1a , ed .
hasselblat , katok , north - holland , 2002 .
knill o 1998 _ elem .
_ * 53 * 89106 lazutkin v f 1973 _ math .
ussr izvestija _ * 7 * 185214 markarian r. 1988 _ comm .
* 118 * 8797 .
markarian r , kamphorst s o and pinto - de - carvalho s 1996 _ comm . math .
phys . _ * 174 * 661679 markarian r , pujals e r , sambarino m : pinball billiards with dominated splitting , etds , 30 , 1757 - 1786 , 2009 .
sys . _ pujals e
r and sambarino m 2009 _ ann .
math . _ * 169 * 675740 tabachnikov s 1995 _ billiards _ ( paris : socit mathmatique de france ) wojtkowski m 1985 _ ergod .
th . & dynam .
_ * 5 * 145161 .
wojtkowski m 1986 _ commun .
phys . _ * 105 * , 391414 .
wojtkowski m 2001 _ smooth ergodic theory and its applications , proc .
pure math .
_ vol 69 ( ams ) 499521 | inspired by the work of pujals and sambarino on dominated splitting @xcite , we present billiards with a modified reflection law which constitute simple examples of dynamical systems with limit sets having dominated splitting and where the dynamics is a rational or irrational rotation . |
over the past several decades , observations of magnetic activity in stars of spectral type f m have uncovered a variety of correlations between the different activity indicators , as well as between the level of activity and properties such as stellar rotation and age .
coronal x - ray emission , transition region uv emission , chromospheric h@xmath0 emission , and non - thermal radio emission increase with both rotation and youth ( e.g. , @xcite ) , and are temporally and energetically correlated in quiescence and during flares ( e.g. , @xcite ) .
these observations have led to a general paradigm of magnetic field amplification at the shearing interface between the radiative and convective zones the so - called @xmath10 dynamo @xcite .
this dynamo operates through a combination of stretching by differential rotation ( @xmath11 ) and twisting by convective motions ( @xmath0 ) . in quiescence ,
the dynamo - generated fields provide a heating source for the chromospheres and coronae primarily through magnetic waves and field dissipation on small scales ( e.g. , @xcite ) .
the interplay between the chromosphere , transition region , and corona is not fully understood , but is known to involve a complex combination of radiation , conduction , and mass flows . in dme stars it has been proposed that the transition region and chromosphere are instead heated by coronal x - rays , leading to the observed typical luminosity ratios of @xmath12 and @xmath13 , respectively ( e.g. , @xcite ) .
in addition to quiescent emission , sudden and large - scale dissipation of the field through large - scale magnetic reconnection may lead to particle acceleration and evaporation of the lower atmosphere into the chromosphere , transition region , and corona , giving rise to correlated radio , x - ray , uv , and h@xmath0 flares ( e.g. , @xcite ) .
the level of activity increases with faster rotation , but eventually saturates at @xmath14 and @xmath15 for rotation periods of @xmath16 d in f
it remains unclear whether this trend reflects a saturation of the dynamo itself or secondary centrifugal effects such as coronal stripping or sweeping of the field toward the poles ( e.g. , @xcite ) .
these activity trends continue to hold even beyond spectral type @xmath17 , where the stellar interiors become fully convective and the @xmath10 dynamo can no longer operate .
indeed , the level of x - ray and h@xmath0 activity peaks in mid m dwarfs at saturated levels @xcite
. this suggests that whatever dynamo mechanism operates in these low mass stars , it is already present in higher mass objects , and becomes increasingly dominant beyond spectral type m3 .
however , the level of x - ray and h@xmath0 activity drops precipitously beyond spectral type m7 , reaching mostly non - detectable levels by spectral type l5 .
this decrease is accompanied by a clear transition from quiescent emission to flares in the few percent of active objects @xcite .
these trends point to a change in the dynamo mechanism , the field configuration , and/or the field dissipation process .
radio observations , however , have uncovered a substantial fraction of active late m and l dwarfs ( @xmath18 or higher ; @xcite ) , which exhibit both quiescent and flaring emission @xcite . unlike the trend in h@xmath0 and x - rays , the level of radio activity appears to _ increase _ with later spectral type @xcite .
the radio emission is orders of magnitude brighter than expected based on the radio / x - ray correlations that are observed in stars down to spectral type m6 , and requires magnetic fields of @xmath19 kg with order unity filling factor sustained over timescales of at least several years @xcite .
thus , contrary to evidence from x - rays and h@xmath0 , a substantial fraction of ultracool dwarfs continue to generate and dissipate magnetic fields . in order to investigate the field generation and dissipation in detail , we have recently initiated a program of simultaneous , multi - wavelength observations of ultracool dwarfs .
such observations are required to trace the temporal evolution of flares across the corona , transition region , and chromosphere , and to study the relation between particle acceleration and heating , particularly in the context of the known correlations in f
m stars . in a previous paper ( @xcite ; hereafter paper i ) we presented observations of the m8.5 rapid rotator , which exhibited a wide range of temporally uncorrelated emission in the radio , h@xmath0 , and x - rays . these included quiescent radio emission from a large - scale field , radio flares from a tangled field component with @xmath20 kg , and periodic h@xmath0 emission matching the stellar rotation period with an inferred hot spot covering about @xmath21 of the stellar photosphere . in addition , the quiescent h@xmath0 emission exceeded the x - ray luminosity by about a factor of two , likely ruling out coronal x - ray emission as the source of chromospheric heating .
here we present observations of the m8 and m8.5 dwarfs and , both of which are known to exhibit magnetic activity .
we find substantially different behavior in each of these two objects , with correlated x - ray / uv flares and quiescent emission in , and uncorrelated radio / h@xmath0 emission in .
along with , the mixed behavior in late m dwarfs thus indicates a transition in the properties of the magnetic field and its dissipation in this spectral type range .
we show that rotation may play at least a partial role in explaining these trends .
we chose to target the nearby dwarf stars and due to previous detections of activity from these objects in the optical / uv / x - ray and radio bands , respectively .
( 8.5 ) is located at a distance of 5.7 pc and has a bolometric luminosity of @xmath22 l@xmath23 @xcite .
a previous @xmath24-hr radio observation revealed persistent emission with @xmath25 @xmath26jy and a limit of @xmath27 on the fraction of circular polarization @xcite .
the rotation velocity of was not previously known , but we measure it here to be @xmath28 km s@xmath9 ( [ sec : optical ] ) , similar to fast rotators such as and kelu-1 .
( m8 ) is located at a distance of 6.1 pc @xcite , has @xmath29 l@xmath23 , and a slow rotation velocity , @xmath30 km s@xmath9 @xcite .
an x - ray flare from was previously detected by rosat , with a duration of @xmath31 min and an average luminosity , @xmath32 erg s@xmath9 , or @xmath33 @xcite .
subsequent chandra observations revealed quiescent emission with @xmath34 @xcite . in addition , _ hubble space telescope _
observations revealed flaring and quiescent transition region emission @xcite . is also known to produce h@xmath0 emission with equivalent widths of @xmath35 reported in the literature @xcite , or @xmath36 .
no radio emission has been detected to date , with a limit of @xmath37 @xmath26jy at 8.5 ghz @xcite .
finally , @xcite recently estimated the surface magnetic field to be @xmath38 kg , where @xmath39 is the field filling factor .
the simultaneous observations of presented here were conducted on 2007 may 3 for a total of 8.7 hr in the radio ( 06:4015:20 ut ) , 8.4 hr in the x - rays ( 07:1615:43 ut ) , and 5.4 hr in the optical ( 09:4215:04 ut ) . observations with the /uvot took place intermittently between 08:09 and 16:31 ut with a total on - source exposure time of 9.4 ks .
observations of were conducted on 2007 july 1 for a total of 8.9 hr in the x - rays ( 10:2919:21 ut ) , with intermittent uv coverage between 11:08 and 19:15 ut for a total of 8.4 ks .
optical spectroscopy was obtained only two weeks later , on 2007 july 15 ( 08:1013:33 ut ) .
very large array observations of were obtained at a frequency of 8.46 ghz in the standard continuum mode with @xmath40 mhz contiguous bands .
scans of 295 s on source were interleaved with 50 s scans on the phase calibrator j1850 + 284 .
the flux density scale was determined using the extragalactic source 3c286 ( j1331 + 305 ) .
data reduction and analysis follow the procedure outlined in paper i. the total intensity and circular polarization light curves are shown in figure [ fig : lsrall ] .
observations were performed with the chandra / acis - s3 backside - illuminated chip , with both and offset from the on - axis focal point by @xmath41 .
a total of 28.33 and 29.29 ks were obtained , respectively .
the data were analyzed using ciao version 3.4 , and counts were extracted in a @xmath42 radius circle centered on the position of each source . for
we established a bore - sight correction based on one source in common between the vla and chandra images .
the derived correction is @xmath43 and @xmath44 . within the @xmath42 aperture centered on the position of we find only 2 counts , with 1.5 counts expected from the background as determined from annuli centered on the source position .
thus , the resulting upper limit is about 7 counts ( @xmath45 confidence level ) .
using an energy conversion factor of @xmath46 erg @xmath47 s@xmath9 ( appropriate for a 1 kev raymond - smith plasma model in the @xmath48 kev range ) we find @xmath49 erg @xmath47 s@xmath9 , or @xmath50 , one of the faintest limits to date for any ultracool dwarf . in the observation of
we detect a total of 60 counts within the @xmath42 extraction aperture ; 2 counts are expected from the background .
of the 60 detected counts , 59 are in the energy range @xmath48 kev , similar to what has been found for other m dwarfs ( including : paper i ) . using this energy range
we construct light curves and spectra .
the light curve is shown at various time binnings in figure [ fig : vb10all ] , and is composed of a bright flare with a duration of @xmath51 hr , followed by about 4 hr of low - level quiescent emission , and finally a fainter flare during the final hour of the observation . using the time range 14:3018:00 ut to represent the quiescent component we find a total of 10 counts , or a count rate of @xmath52 s@xmath9 . to determine the flux and plasma temperature we fit the @xmath48 kev spectrum using a raymond - smith model .
we find that a single - component model provides a poor fit to the data , with @xmath53 for 9 degrees of freedom ( null hypothesis probability of 0.2 ) ; figure [ fig : vb10xspec ] .
the fit parameters are @xmath54 kev , and a normalization of @xmath55 .
an improved fit is achieved using two components , with @xmath56 for 7 degrees of freedom ( null hypothesis probability of 0.94 ) ; figure [ fig : vb10xspec ] . the best - fit temperatures are @xmath57 kev and @xmath58 kev ( formally , @xmath59 kev ) .
the normalizations of the two components are @xmath60 and @xmath61 , respectively .
the resulting average flux over the full observation is @xmath62 erg @xmath47 s@xmath9 , or @xmath63 . using the same model for the quiescent component
, we find @xmath64 erg @xmath47 s@xmath9 , or @xmath65 , in excellent agreement with the value found previously by @xcite .
the peak count rate of the first flare is about @xmath66 s@xmath9 , corresponding to @xmath67 erg @xmath47 s@xmath9 and @xmath68 . for the second flare , @xmath69 erg @xmath47 s@xmath9 and @xmath70 .
we used the gemini multi - object spectrograph ( gmos ; @xcite ) mounted on the gemini - north 8-m telescope with the b600 grating set at a central wavelength of 5250 , and with a @xmath71 slit .
the individual 300-s exposures were reduced using the gemini package in iraf ( bias subtraction , flat - fielding , and sky subtraction ) .
wavelength calibration was performed using cuar arc lamps and air - to - vacuum corrections were applied .
the spectra cover @xmath72 at a resolution of about 5 . for we obtained a series of sixty exposures , while for the data were obtained non - simultaneously with a total of 61 exposures .
the fast readout time of @xmath73 s provides @xmath74 on - source efficiency .
we detect highly variable h@xmath0 emission in ( figure [ fig : lsrall ] ) .
spectra with strong h@xmath0 emission also exhibit higher - order balmer lines ( h@xmath75h@xmath76 ) and h&k emission .
sample spectra in the high and low emission line states are shown in figure [ fig : lsrspec ] .
the spectrum of exhibits strong variability in the balmer ( h@xmath0h@xmath77 ) and h&k lines ; figure [ fig : vb10balmer ] . as in the case of
we identify high and low emission line states ( figure [ fig : vb10spec ] ) , but we also identify an impulsive flare state marked by a short duration ( @xmath78 s ; figure [ fig : vb10balmer ] ) and emission lines of ( figure [ fig : vb10spec ] ) .
in addition to the low - resolution gemini spectra , we obtained high resolution spectra of to measure its rotation velocity , using the high resolution echelle spectrometer ( hires ) on the keck i 10-m telescope .
the spectroscopic setup and data reduction are detailed in @xcite and @xcite . using the slow rotator cn leo ( m6 ; @xmath79 km s@xmath9 ) as a template , we measure a rotation velocity for of @xmath28 km s@xmath9 , or a period of about 2.4 hr ; figure [ fig : lsrrot ] .
data were obtained with the uv / optical telescope in the uvw1 filter ( @xmath80 ) , as a series of 6 images with exposure times ranging from 560 to 1630 s for ( figure [ fig : lsrall ] ) and 980 to 1630 s for ( table [ tab : vb10uvot ] and figure [ fig : vb10all ] ) .
no source is detected at the position of in any of the individual exposures , but we detect a source at @xmath81 confidence level in the combined 9.4 ks image .
photometry of the combined image ( including a 0.2 mag correction to the standard @xmath82 aperture ) reveals a flux of @xmath83 erg @xmath47 s@xmath9 @xmath9 , or @xmath84 mag . to estimate the expected uv photospheric emission we convolve the uvot / uvw1 transmission curve with an ames - cond model with @xmath85 and @xmath86 k @xcite , which provides an excellent fit to the optical and ir data for ( figure [ fig : lsrsed ] ) .
we find an expected photospheric brightness of @xmath87 mag , in excellent agreement with the observed flux .
thus , no contribution from a quiescent or flaring transition region is detected .
, on the other hand , is clearly variable in the uv , and is detected in exposures 1 , 2 , and 6 , with an additional detection when combining exposures 3 , 4 , and 5 .
a summary of the derived magnitudes , corrected to the standard @xmath82 aperture , is provided in table [ tab : vb10uvot ] , and the light curve is shown in figure [ fig : vb10all ] . repeating the same analysis as for , we find that the minimum observed uv emission exceeds the expected photospheric emission by about an order of magnitude .
thus , in the flaring and quiescent uv emission are due to an active transition region .
we observed and across a wide wavelength range that traces activity in various layers of the outer atmosphere .
the radio emission is due to particle acceleration by magnetic processes , the optical emission lines trace the chromospheric plasma , the uv emission arises in the transition region , and the x - ray thermal emission is produced in the corona . [ [ sec : lsr ] ] the broad - band emission properties of are summarized in figure [ fig : lsrall ] .
the radio emission appears to be nearly constant both in total intensity and circular polarization .
the average flux density is @xmath88 @xmath26jy , while the @xmath89 upper limit on the fraction of circular polarization is @xmath90 . for comparison
, previous observations of revealed a flux of @xmath25 @xmath26jy , with an upper limit on the circular polarization of @xmath91 @xcite .
we note the possible detection of one weak flare ( at 08:05 ut ) with a peak flux density of about 1.3 mjy and a duration of about 5 min ( figure [ fig : lsrall ] ) .
the average flux density during the putative flare is only @xmath92 @xmath26jy above the quiescent emission level , and the average fraction of circular polarization is @xmath93 .
we therefore conclude that the radio emission from is generally persistent in origin , with at most a mild variability at the level of @xmath94 on hour to year timescales .
the one possible short duration flare points to a duty cycle of less than a few percent if the flare distribution is uniform .
alternatively , it is possible that we observed in a period of relative quiescence , and future observations may reveal the existence of flares with a higher duty cycle . since our radio observation covers nearly 4 rotation periods of , the lack of significant variability points to emission from a uniform and stable magnetic field configuration . assuming a structure with @xmath95 cm we infer a brightness temperature , @xmath96 k , indicative of non - thermal gyrosynchrotron emission .
this conclusion is supported by the overall persistence of the radio emission and the low fraction of circular polarization . in the context of gyrosynchrotron radiation the emission spectrum
is determined by the size of the emitting region ( @xmath97 ) , the density of radiating electrons ( @xmath98 ) , and the magnetic field strength ( @xmath99 ) according to @xcite : @xmath100 @xmath101 @xmath102 where @xmath103 is the angle between the magnetic field and the line of sight . using a range of @xmath104 we find @xmath105 g , respectively .
assuming @xmath106 ghz , we find @xmath107 cm , and @xmath108 @xmath109 ; the latter is for the range @xmath110 .
typical coronal densities in m dwarfs are @xmath111 @xmath109 @xcite , suggesting that @xmath112 , and hence @xmath113 g and @xmath114 . since the corona is likely to have an overall shell geometry , the derived physical properties can be interpreted as a structure extending @xmath115 above the stellar photosphere with order unity filling factor .
smaller filling factors will lead to more extended magnetic structures . with strong persistent radio emission and no corresponding x - ray emission ,
joins the sample of ultracool dwarfs which strongly violate the radio / x - ray correlation , @xmath116 hz@xmath9 @xcite , observed in f
m6 active stars and in solar flares .
we find @xmath117 hz@xmath9 , more than four orders of magnitude in excess of the expected value ; figure [ fig : gb ] .
we stress that the long - term persistence and low circular polarization of the radio emission clearly point to gyrosynchrotron radiation from a stellar - scale magnetic field , so the severe violation can not be explained away with small - scale , short - duration heating by coherent processes such as electron cyclotron maser or plasma radiation .
we next turn to the h@xmath0 emission for which we measure a range of equivalent widths of @xmath118 ( figure [ fig : lsrall ] ) .
this range corresponds to @xmath119 , about an order of magnitude lower than the saturated h@xmath0 emission found in mid m dwarfs .
the weaker chromospheric emission from is in good agreement with the general trend of decreasing h@xmath0 emission in ultracool dwarfs . as in the case of the radio emission , the level of detected h@xmath0 emission is puzzling in the context of the non - detected x - ray emission .
observations of m0m6 dwarfs reveal a typical ratio , @xmath120 , with a full range of @xmath121 @xcite . for
this ratio is @xmath122 ( using the weakest h@xmath0 flux ) , at least an order of magnitude larger than in m0m6 dwarfs .
the significantly brighter h@xmath0 emission indicates that chromospheric heating is significantly more efficient in than coronal heating , possibly as a result of the field configuration .
we return to this point in [ sec : rot ] .
moreover , the elevated h@xmath0 emission rules out coronal x - ray emission as the source of chromospheric heating , as proposed for dme stars . while both radio and h@xmath0 emission are detected , we find a clear lack of temporal correlation between these two activity indicators .
the h@xmath0 light curve exhibits significant variability on timescales of @xmath1 hr , with a wide range of amplitudes , @xmath123 .
however , no clear radio variability is observed in coincidence with any of the h@xmath0 brightening episodes . in particular , during the large increase in h@xmath0 equivalent width centered on 13:40 ut ( figure [ fig : lsrall ] ) , we find a limit of @xmath124 on the variability of the radio emission .
this can be explained in two ways .
first , the h@xmath0 variability originates from much smaller physical scales than the radio emission ( for which @xmath125 ) , and any associated radio variability is therefore dwarfed by the overall persistent flux . in this case , we infer that the typical scale of chromospheric regions giving rise to h@xmath0 variability is @xmath6 of the overall chromospheric volume . alternatively , the radio emission associated with the variable h@xmath0 emission is produced at a much different frequency than our observed 8.5 ghz band
. this may be the result of coherent radio emission from fields weaker than @xmath51 kg , which would lead to emission at @xmath126 hz . to summarize , exhibits quiescent radio emission with no appreciable variability over the timescale of our observation ( 4 rotation periods ) , or on year timescales .
the chromospheric h@xmath0 emission , on the other hand , is highly variable with changes of a factor of 3 in equivalent width on timescales of @xmath127 hr . the h@xmath0 flux exceeds the undetected x - ray emission by at least an order of magnitude , indicating that the chromosphere is not heated by coronal emission .
the radio emission violates the radio / x - ray correlation of f
m6 stars by about 4 order of magnitude .
[ [ sec : vb ] ] for we were able to obtain simultaneous observations only in the x - rays and uv ( figure [ fig : vb10all ] ) .
the x - ray light curve exhibits both flaring and quiescent emission with ratios of @xmath128 and @xmath129 , respectively . as noted in
[ sec : obs ] , quiescent x - ray emission was previously detected from in a shorter observation @xcite and we confirm its existence here at the same flux level . the x - ray spectrum is dominated by a @xmath130 k plasma , with a possible second component with @xmath131 k ( figure [ fig : vb10xspec ] ) .
these temperatures are typical of coronal x - ray emission from m dwarfs , and similar to what we previously found for , @xmath132 k ( paper i ) .
the uv emission clearly tracks the x - ray behavior , with detections of the two flares and the quiescent component .
the peak uv brightness exceeds the quiescent emission by a factor of about 5 ( table [ tab : vb10uvot ] ) , somewhat less that the order of magnitude change in x - ray brightness ( [ sec : xrays ] ) .
moreover , the quiescent component exceeds the expected photospheric emission by about an order of magnitude , and thus points to the existence of a persistent transition region , as noted previously by @xcite .
so far this is the only case of correlated emission found in any of our targets , suggesting that at least some late m dwarfs may follow the behavior seen in early m dwarfs . as in the case of the x - ray and uv data
, our non - simultaneous optical spectroscopy also reveals a significant level of variability .
as far as we know , our spectroscopic observations provide the most extensive coverage of in the published literature , and are thus valuable even if no multi - wavelength coverage is available . here
we provide a brief summary of the observations , and defer a detailed analysis to a future publication .
we clearly detect h@xmath0 in emission in all of the individual exposures .
an equivalent width light curve is shown in figure [ fig : vb10balmer ] .
the range of measured values is @xmath133 , extending to somewhat higher values than previously published for .
similarly , the h@xmath75 line varies from 5 to 26 , while h@xmath134 and h@xmath76 range from undetectable levels ( @xmath135 ) to about 80 .
we also detect h&k emission ranging from non - detectable levels to a k line equivalent width of about 30 .
the emission line light curves exhibit a wide range of variability timescales .
we note that the total duration of our observation samples only about 1/4 of a rotation of ( @xmath136 km s@xmath9 ) , and we therefore have no information on rotational modulation , which was detected in our previous target , ( paper i ) .
the variability is primarily gradual with the rise and fall times of individual episodes being roughly the same .
the broadest brightening has a rise time of about 0.7 hr and a decay time of about 2 hr , with a factor of two increase in equivalent width .
the variability is reflected in all of the detected balmer lines , but the relative change in equivalent width is largest for the weakest lines . this mild change in the balmer decrement points to increased densities and temperatures during brighter emission episodes .
sample spectra in these high and low states are shown in figure [ fig : vb10spec ] .
in addition to the time - resolved and gradual changes in emission line equivalent width we detect two episodes of rapid brightening with a duration of @xmath78 s ( flare state ) .
these flares are marked by a large increase in the equivalent width of the higher - order balmer lines and the lines relative to the high state spectra .
they are further distinguished from the high state spectra by the appearance of lines ( figure [ fig : vb10spec ] ) .
temporally , both of the detected flares are superposed on gradual brightenings , suggesting that the two modes of variability are causally and temporally unrelated and arise in different regions of the chromosphere .
the significant difference in balmer decrement between the high and flare states , and the appearance of lines in the latter indicate higher plasma densities , and most likely temperatures in the flare state . a comparison to flares on mid m dwarfs ( @xcite and
references therein ) reveals shallower balmer decrements in the impulsive flares on : @xmath137 , @xmath138 , and @xmath139 , compared to average values of 0.8 , 0.5 , and 0.4 , respectively .
similarly , the formation of lines generally requires dense plasma ( @xmath140 @xmath109 ) at @xmath141 k , approaching the transition region temperatures .
although our x - ray and optical observations are not simultaneous , the variability timescales in both bands are similar ( figures [ fig : vb10all ] and [ fig : vb10balmer ] ) .
this suggests that in chromospheric heating by coronal x - rays may in fact work .
indeed , the observed range of @xmath142 ( table [ tab : vb10spec ] ) is similar to the range of @xmath143 .
thus , the chromospheric and coronal structure of is markedly different than that of , for which the h@xmath0 luminosity exceeds the x - ray luminosity by at least an order of magnitude , and no evidence for plasma at coronal temperatures exists .
to conclude , the magnetic activity in differs significantly from that of and , and is more typical of m0m6 dwarfs . the x - ray and uv emission
are clearly correlated and point to the presence of quiescent and flaring corona and transition region .
the optical chromospheric emission exhibits similar timescales and amplitudes of variability , and can be separated into a gradually variable component ( low and high states ) , and a rapidly flaring component ( flare state ) that is marked by higher plasma densities and temperatures .
with the increasing sample of ultracool dwarfs observed in the radio and x - rays we can begin to address the physical conditions that underlie the generation of magnetic fields and their influence on the coronae and chromospheres .
it is clear from our work that late m dwarfs exhibit a mix of activity patterns in comparison to stars with spectral types f to early m stars , whose magnetic activity is powered by the @xmath10 dynamo .
thus , it is instructive to investigate whether rotation also plays a role in the magnetic activity of ultracool dwarfs . the generation of persistent x - ray emission in f
m stars is still not fully understood , but its correlation with radio activity and rotation velocity suggests that it is directly tied to heating by magnetic field dissipation and particle acceleration . in figure
[ fig : lxrot ] we show the well - known rotation - activity relation and saturation of @xmath144 for stars with spectral type f m6 .
the rotation - activity relation provides evidence for the existence of an @xmath10 dynamo , while three primary explanations have been proposed for the saturation of @xmath145 ( see also @xcite ) .
the first is saturation of the dynamo mechanism itself at low rossby numbers , @xmath146 ( e.g. , @xcite ) ; here @xmath147 is the convective turnover time .
second , centrifugal effects in rapid rotators may lead to a circulation pattern that sweeps the magnetic fields from the lower convection zone toward the poles , thereby reducing the field filling factor @xcite .
the third explanation is centrifugal stripping of the corona at the corotation radius , @xmath148 @xcite . in this scenario , the particle density , @xmath149 , and emission volume , @xmath150 , conspire to produce an emission measure ( @xmath151 ) that is independent of the rotation period , i.e. , saturation .
however , this balance breaks down when the corotation radius is well inside the corona , @xmath152 , leading to super - saturation a reduction in @xmath144 at very rapid rotation rates ( e.g. , @xcite ) .
it is instructive to study whether the x - ray emission of ultracool dwarfs exhibits saturation and/or super - saturation , and if so , which of the mechanisms described above is responsible for these effects .
the values of @xmath144 as a function of rotation period presented here and in paper i , as well as in the literature , are summarized in figure [ fig : lxrot ] .
two trends are clear . in the range of rotation periods of @xmath153 d ,
in which f m6 stars exhibit saturated emission , @xmath154 for the ultracool dwarfs is a few times smaller than the saturated value of the f m6 stars .
this points to generally weaker coronae in ultracool dwarfs compared to higher mass stars , although in one case an unusually bright flare with @xmath155 was detected @xcite . on the other hand
, ultracool dwarfs with periods of @xmath156 d ( @xmath157 km s@xmath9 ) exhibit much lower values of @xmath158 .
these include our targets and .
the reduction in x - ray activity at high rotation velocities is reminiscent of the super - saturation phenomenon . in the context of centrifugal stripping , an ultracool dwarf with @xmath159 m@xmath23 and @xmath160 r@xmath23
has a corotation radius that is smaller than the coronal radius ( @xmath161 above the photosphere ) when @xmath162 hr .
this is a factor of two times smaller than the fastest rotators in the sample , with @xmath163 km s@xmath9 .
this indicates that centrifugal stripping may play a role in reducing @xmath144 for fast rotators only if the the magnetic field is generally dominated by extended loops with @xmath164 , which may indeed be the relevant scale for m dwarfs ( e.g. , @xcite ) . in the case of , ( paper i ) , and 2massj00361617 + 1821104 @xcite we inferred from our radio observations a magnetic field scale of @xmath165 cm , in the rough range required for efficient centrifugal stripping .
thus , to the extent that existing observations of ultracool dwarfs reveal possible super - saturation in the x - rays , it is presently unclear whether this is due to intrinsic dynamo effects or to the secondary effect of centrifugal stripping .
the proposed mechanism of centrifugal clearing of the equatorial plane in rapid rotators @xcite is not likely to play a role since the persistent radio emission on @xmath166 timescales in several ultracool dwarfs points to fields with order unity covering fractions .
an alternative possibility for the drop in @xmath144 is that the source of coronal heating is different , and much less efficient , than in higher mass stars .
this scenario may explain the lower values of @xmath144 in the regime where f m6 stars are saturated .
however , it is unclear why such a scenario should depend on rotation , particularly since the radio emission ( i.e. , particle acceleration ) does not diminish in fast rotators ( see [ sec : radio ] ) .
we caution that the possibility of super - saturation in the x - rays is based on a small number of objects , and moreover that the effect of centrifugal stripping depends on the magnetic field configuration .
future x - ray observations of additional late m and l dwarfs , as well as the determination of rotation velocities for a larger sample of ultracool dwarfs will reveal whether , and at what rotation velocity , super - saturation may set in present .
if confirmed , the value of @xmath167 at which super - saturation occurs may hold a clue to the nature of the dynamo or the dominant physical scale of coronal magnetic loops .
since radio emission is arguably a more robust tracer of magnetic activity than x - ray emission , we repeat the same exercise for @xmath168 as a function of rotation period and velocity . @xcite noted a possible correlation between radio activity and rotation based on a small sample of ultracool dwarfs , and we re - visit this investigation here . in figure
[ fig : lrrot ] we plot radio detections and upper limits for objects with a known rotation velocity and period .
we find that for objects with spectral type @xmath169 those with @xmath170 km s@xmath9 , or @xmath171 hr , exhibit a typical quiescent radio activity , @xmath172 .
on the other hand , those with @xmath173 km s@xmath9 have @xmath174 , nearly an order of magnitude larger .
if we include objects earlier than m7 , the increase in radio activity with rotation velocity becomes even more pronounced , with a typical level of @xmath175 for @xmath176 km s@xmath9 .
the majority of the ultracool dwarfs with a known rotation velocity remain undetected in vla observations with a typical duration of @xmath177 hr , mostly as a result of the decrease in @xmath178 for l and t dwarfs ( figure [ fig : lrrot ] and @xcite ) . still , it appears that with the present sample , there is possible evidence for a radio rotation - activity relation , with no evidence for either saturation or super - saturation .
this is contrary to the apparent decrease in @xmath144 at @xmath179 km s@xmath9 ( [ sec : xr ] ) .
if confirmed with future observations , the most likely explanation is that the dynamo mechanism is at least partially driven by rotation ( even in the presence of full convection ) , with no apparent saturation at very low rossby numbers ( @xmath180 ) , contrary to the trend in f m6 stars .
regardless of the exact dynamo mechanism there is a clear discrepancy between the radio and x - ray observations in terms of a possible activity - rotation relation and its super - saturation .
the lack of clear saturation or super - saturation in the radio indicates that the dynamo itself is not likely to saturate even at high rotation velocities .
the possible decline in x - ray activity at @xmath8 km s@xmath9 may thus be the result of secondary effects such as inefficient heating of the plasma to coronal temperatures , or centrifugal stripping of the most extended magnetic loops , with @xmath181 .
the former scenario is supported by the observed ratio @xmath182 in the quiescent emission from and , which points to heating of chromospheric plasma by a process other than coronal x - ray emission .
the difference in trends between radio and x - ray activity as a function of rotation velocity may also underlie the violation of the radio / x - ray correlation in ultracool dwarfs ( figure [ fig : gb ] ) . in figure
[ fig : bgrot ] we plot @xmath183 relative to its standard value in f
m6 stars ( @xmath184 ; @xcite ) as a function of rotation velocity .
we find that objects with @xmath173 km s@xmath9 exhibit a significantly more severe violation of the radio / x - ray correlation than those with slower rotation .
it is important to note , however , that rotation alone may not fully account for the excess ratios since even objects with @xmath185 km s@xmath9 may exceed the value of @xmath186 by an order of magnitude or more .
still , it appears that faster rotators generally violate the radio / x - ray correlation by a larger factor .
we presented simultaneous multi - wavelength observations of two late m dwarfs to trace the magnetic activity in their outer atmospheres . in the case of ( m8.5 )
we detect persistent radio emission and highly variable h@xmath0 emission . no excess uv emission is detected .
similarly , we detect no x - ray emission to a deep limit of @xmath2 .
the ratio of radio to x - ray flux exceeds the average value measured in a wide range of active stars by more than 4 orders of magnitude .
similarly , the ratio @xmath187 exceeds the values measured for m0m6 dwarfs by at least an order of magnitude , and rules out heating of the quiescent chromosphere by coronal emission .
temporally , we find no correspondence between the radio and h@xmath0 light curves .
this indicates that even if the source of chromospheric heating is magnetic reconnection , it occurs on sufficiently small scales that the overall radio emission does not change by more than @xmath18 .
alternatively , the radio emission produced in conjunction with the h@xmath0 variability may peak at a lower frequency than our 8.5 ghz observations .
in the former scenario , our limit on the radio variability during the brightest h@xmath0 episode can be interpreted as a limit of @xmath6 on the chromospheric volume involved in the variable h@xmath0 emission . in the latter scenario ,
it is possible that the coincident radio emission is dominated by coherent emission at @xmath188 , with @xmath189 kg . with our measured rotation velocity of @xmath190 km s@xmath9 for , the radio observations sample nearly 4 full rotations .
the stability of the radio emission thus requires a large - scale and uniform magnetic field , with an inferred strength of @xmath5 g. the similarity in flux to a past observation suggests that the field is stable on year timescales .
this may be expected given the long convective turnover time for ultracool dwarfs , on the order of several years .
, on the other hand , exhibits bright and variable x - ray and uv emission , including a pair of flares and quiescent emission . the behavior in both bands , which trace the corona and transition region , respectively , is highly correlated . from the x - ray spectrum
we infer a coronal temperature of about @xmath191 k , with a possible second component at @xmath192 k. the optical emission lines exhibit a similar amplitude and timescale of variability to the x - rays and uv , with two distinct states of gradual and impulsive flaring . the lack of causal relation between the two states points to emission from distinct plasma environments . moreover ,
the shallow balmer decrement and emission in the impulsive flare state require significantly denser and hotter plasma .
our detailed study of , , and reveals mixed patterns of of behavior compared to the magnetic activity in f - m6 stars , and thus points to a transition in the atmospheric structure and heating process in the late m dwarf regime .
in particular , stellar - scale magnetic fields are present , as evidence by quiescent and uniform radio emission , but the coronae are generally weaker than in early m dwarfs .
the chromospheric activity declines less rapidly , and all three targets exhibit a similar range of highly variable h@xmath0 emission , @xmath193 .
this range is about an order of magnitude less that the saturated value of mid m dwarfs . in the standard picture of solar and stellar magnetic flares , the release of magnetic energy through processes such as reconnection leads to acceleration of electrons , and subsequently heating of the chromosphere , transition region , and corona through evaporation of the lower atmosphere . in quiescence ,
the chromospheres of m0m6 dwarfs may instead be heated by coronal x - ray emission @xcite , as evidenced by the typical observed ratios of @xmath194 @xcite .
such a mechanism is clearly not at play in the case of , and most likely , since the chromospheric emission is significantly more energetic than the non - detected corona .
it is possible instead that the source of chromospheric heating is continuous micro - flaring activity , leading to replenishment of chromospheric and transition region plasma by evaporation . in this scenario
the weak or absent coronal emission may be due to a limited temperature enhancement of @xmath195 k , which is sufficient to produce chromospheric emission , but no significant soft x - ray emission .
the continuously variable h@xmath0 emission in appears to support this idea , and along with the lack of corresponding radio variability points to heating on scales much smaller than the stellar photosphere . as we noted in
[ sec : rot ] , it is also possible that rapid rotation in ultracool dwarfs suppresses the x - ray emission through centrifugal stripping . the apparent increase in radio activity with rotation indicates that the dynamo itself does not saturate , at least up to @xmath196 km s@xmath9 , which is roughly @xmath197 of the break - up velocity .
the effects leading to the apparent super - saturation in the x - rays may also underlie the severe violation of the radio / x - ray correlation since the level of violation appears to be correlated with rotation velocity ( figure [ fig : bgrot ] ) .
we stress that the role of rotation is still speculative due to the small number of objects with x - ray , radio , and rotation measurements .
it is therefore crucial to increase the sample of ultracool dwarfs for which these quantities are measured . to summarize ,
we conclude that late m dwarfs mark a change in the properties of the magnetic field and its dissipation , as well as the generation of high temperature plasma in the outer atmosphere .
the weak x - ray emission , both in relation to @xmath178 and @xmath198 , points to inefficient heating of plasma to coronal temperatures , or alternatively to stripping of the most extended magnetic loops .
the lack of temporal correlation between the radio and h@xmath0 activity , however , indicates that even if atmospheric evaporation is taking place , it occurs on smaller physical scale than the overall magnetic field structure .
we end by noting that the use of simultaneous , multi - wavelength , and long duration observations to probe the magnetic activity of individual ultracool dwarfs provides a crucial complement to large single - band surveys .
in particular , the long duration and high time resolution data elucidate the range of timescales and amplitudes of gradual and impulsive flares , and can moreover uncover rotationally - induced modulations , as in the case of ( paper i ) .
in addition , the existence or absence of correlations between the various activity indicators ( as in and ) , and their implications for the magnetic field and its dissipation , could not have been inferred from existing observations . in the upcoming chandra cycle
we will expand our analysis with observations of several l0l3 dwarfs .
these observations will reveal whether the transition in magnetic activity patterns seen in the late m dwarfs continues to lower mass objects , some of which are bona - fide brown dwarfs .
we thank the chandra , gemini , vla , and swift schedulers for their invaluable help in coordinating these observations .
this work has made use of the simbad database , operated at cds , strasbourg , france .
it is based in part on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) and conicet ( argentina ) .
data from the uvot instrument on swift were used in this work .
swift is an international observatory developed and operated in the us , uk and italy , and managed by nasa goddard space flight center with operations center at penn state university .
support for this work was provided by the national aeronautics and space administration through chandra award number g07 - 8014a issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics space administration under contract nas8 - 03060 .
lll 0.1 in 1 & 11:08:2211:31:00 & @xmath199 + 2 & 12:32:2312:49:00 & @xmath200 + 3 & 14:08:2714:36:00 & @xmath201 + 4 & 15:44:2816:12:00 & @xmath202 + 5 & 17:21:2717:49:00 & @xmath202 + 35 & 14:08:2717:49:00 & @xmath203 + 6 & 18:57:2719:15:00 & @xmath204 lccccccccc 0.1 in h@xmath0 & 4.3 & 5.2 & @xmath205 & 8.6 & 13.3 & @xmath206 & 9.0 & 15.1 & @xmath207 + h@xmath75 & 4.8 & 0.5 & @xmath208 & 17 & 2.8 & @xmath209 & 33 & 8.7 & @xmath210 + h@xmath134 & 9.3 & 0.2 & @xmath211 & 31 & 1.6 & @xmath212 & 75 & 8.7 & @xmath210 + h@xmath76 & & & & 31 & 1.0 & @xmath213 & 95 & 8.0 & @xmath214 + h@xmath215+h & 8.5 & 0.2 & @xmath211 & 111 & 1.8 & @xmath216 & 83 & 7.8 & @xmath214 + h@xmath77 & & & & 9.1 & 0.3 & @xmath217 & 44 & 4.8 & @xmath205 + k & 15 & 0.2 & @xmath211 & 39 & 1.2 & @xmath218 & 26 & 2.5 & @xmath219 + @xmath220 & & & & & & & 2.8 & 1.1 & @xmath213 + @xmath221 & & & & & & & 5.3 & 0.9 & @xmath213 + @xmath222 & & & & & & & 5.4 & 0.5 & @xmath208 kev ( @xmath53 for 9 d.o.f ) , and _ right : _ a two - component raymond - smith model with @xmath223 kev and @xmath58 kev ( @xmath56 for 7 d.o.f ) .
the single - temperature model leaves significant residuals , which are largely eliminated by the addition of hotter component .
[ fig : vb10xspec],width=350 ] kev ( @xmath53 for 9 d.o.f ) , and _ right : _ a two - component raymond - smith model with @xmath223 kev and @xmath58 kev ( @xmath56 for 7 d.o.f ) .
the single - temperature model leaves significant residuals , which are largely eliminated by the addition of hotter component .
[ fig : vb10xspec],width=350 ] | as part of our on - going investigation of magnetic activity in ultracool dwarfs we present simultaneous radio , x - ray , uv , and optical observations of ( m8.5 ) , and simultaneous x - ray and uv observations of ( m8 ) , both with a duration of about 9 hours .
exhibits persistent radio emission and h@xmath0 variability on timescales of @xmath1 hr . the detected uv flux is consistent with photospheric emission , and no x - ray emission is detected to a deep limit of @xmath2 .
the h@xmath0 and radio emission are temporally uncorrelated , and the ratio of radio to x - ray luminosity exceeds the correlation seen in f
m6 stars by @xmath3 .
similarly , @xmath4 is at least 30 times larger than in early m dwarfs , and eliminates coronal emission as the source of chromospheric heating .
the lack of radio variability during four rotations of requires a uniform stellar - scale field of @xmath5 g , and indicates that the h@xmath0 variability is dominated by much smaller scales , @xmath6 of the chromospheric volume .
, on the other hand , shows correlated flaring and quiescent x - ray and uv emission , similar to the behavior of early m dwarfs .
delayed and densely - sampled optical spectra exhibit a similar range of variability amplitudes and timescales to those seen in the x - rays and uv , with @xmath7 . along with our previous observations of the m8.5 dwarf
we conclude that late m dwarfs exhibit a mix of activity patterns , which points to a transition in the structure and heating of the outer atmosphere by large - scale magnetic fields .
we find that rotation may play a role in generating the fields as evidenced by a tentative correlation between radio activity and rotation velocity .
the x - ray emission , however , shows evidence for super - saturation at @xmath8 km s@xmath9 , which could be the result of secondary effects such as inefficient heating or centrifugal stripping of extended coronal loops . these effects may underlie the severe violation of the radio / x - ray correlation in ultracool dwarfs .
upcoming observations of l dwarfs will reveal whether these trends continue in substellar objects . |
sliding ropes and falling chains are often used in introductory mechanics course as examples of variable mass systems @xcite .
the problems at first seem deceptively simple but at closer inspection they are nt such .
an interesting history of falling chain problem @xcite reveals a considerable amount of both insight and confusion which often pervades such kind of problems .
conceptual difficulties of variable mass systems even was considered @xcite as an argument against their inclusion at all in any introductory mechanics course . however , variable mass systems play such an important role in modern day technology that it is justified even beginning students to be exposed to main principles of the mechanics of the systems of variable composition .
we agree with @xcite that the best way to make the subject more palatable for beginning students is to introduce it through a problem that allows complete enough treatment even without any reference to the variable mass systems and then , after the correct answers are already guessed , illustrate how the main principles of the variable mass systems can illuminate the problem .
in this note we reconsider an old problem @xcite which , in our opinion , is ideally suited for goals to introduce the concept of momentum flux and illustrate its use for systems of variable composition .
a folded heavy rope of mass @xmath0 and length @xmath1 is suspended on a little frictionless peg so that its right arm is a bit longer .
the rope is released and begins to slide down under the action of gravity .
what is the weight of the sliding rope as felt by the peg ? let the length of the right arm of the rope is @xmath2 at some moment of time . when the @xmath2-component of the rope s momentum will be @xmath3 where we have assumed that the @xmath2-axis is directed downward , see figure [ fig1 ] .
we should have @xmath4 where @xmath5 equals just the weight of the sliding rope as felt by the supporting peg . from ( [ eq1 ] ) and ( [ eq2 ] ) we get @xmath6 .
\label{eq3}\ ] ] note that this last equation also follows from @xmath7 where @xmath8=\frac{1}{4l}\left [ z^2+(2l - z)^2\right ] \ ] ] is the center of mass coordinate of the rope .
this is not surprising because for the rope , as a whole , the mass is not changed and therefore both forms of newton s equations , @xmath9 and @xmath10 are equivalent . the velocity of the rope , @xmath11 , can be found from energy conservation @xmath12 which gives @xmath13=\frac{g}{l}(z - l)^2 . \label{eq6}\ ] ] differentiating ( [ eq6 ] ) , we can find the acceleration @xmath14 of the rope @xmath15 substituting ( [ eq6 ] ) and ( [ eq7 ] ) into ( [ eq3 ] ) , we get easily @xmath16
. \label{eq8}\ ] ] so far so good .
however , we can think naively that @xmath17 where @xmath18 is the tension of the rope near the peg .
so ( [ eq9 ] ) opens a second easy way to find the sliding rope s weight .
namely , the equations of motion for the right and left parts of the rope are respectively @xmath19 and @xmath20 if ( [ eq11 ] ) is subtracted from ( [ eq10 ] ) , we just get ( [ eq7 ] ) for the rope s acceleration .
the tension @xmath18 can be found from ( [ eq10 ] ) , for example , @xmath21 .
\label{eq12}\ ] ] and here a big surprise is waiting for us : clearly ( [ eq8 ] ) , ( [ eq9 ] ) and ( [ eq12 ] ) are not compatible !
so what is the correct weight of the rope ?
a first thought is that somehow ( [ eq8 ] ) is incorrect , because it predicts that the weight of the sliding rope vanishes at @xmath22 and beyond that even becomes negative !
however , the truth is that both ( [ eq8 ] ) and ( [ eq12 ] ) are correct , as the logic leading to them was rock - solid . only ( [ eq9 ] )
is at error . to understand why ,
let us trace the fate of the small piece of the rope that disappears from the left arm just to join the right arm after an instant .
the `` instant '' , however , has a finite duration @xmath23 if the peg s radius is @xmath24 so that the length of the fold of the rope is @xmath25 . during @xmath26 ,
a piece of the length @xmath25 disappears from the left arm of the rope and an equivalent piece of the same length joins the right arm .
however , the piece that disappeared from the left arm had upward velocity @xmath27 , while the velocity @xmath11 of the part that joins the right arm points downward .
therefore , a small mass @xmath28 , which was transferred from one sub - rope to the other , experiences an effective downward acceleration @xmath29 when the radius @xmath24 of the peg goes to zero , @xmath30 increases without bound .
however , @xmath31 goes to zero as well , so that @xmath32 remains finite : @xmath33 therefore , from @xmath34 we get @xmath35 which is just the relation between ( [ eq8 ] ) and ( [ eq12 ] ) . to treat the problem more systematically ,
this is a good point to introduce variable mass systems and indicate that the root of our confusion , which led to ( [ eq9 ] ) , can be traced to the fact that for variable mass systems neither ( [ eq4 ] ) nor ( [ eq5 ] ) are , in general , valid equations of motion . instead ,
the right starting point for variable mass systems is @xcite @xmath36 where @xmath37 is the momentum flux into the system - how much momentum is brought into the system by new parts in unite time .
the logic of ( [ eq15 ] ) is simple : the momentum of the system of variable composition changes not only because of the action of the resultant external force @xmath38 , but also because new parts of the system can bring some momentum into the system .
left and right arms of the rope can be considered as variable mass systems . in unite time , new parts bring the momentum @xmath39 into the right arm .
parts that disappear from the left arm have negative momentum .
therefore the @xmath2-component of the momentum flux into the left arm is also positive : as a result of disappearance of those parts from the left arm @xmath2-component of its momentum increases ( becomes less negative ) .
then ( [ eq15 ] ) indicates that the equations of motion for the right and left parts of the rope are respectively @xmath40 and @xmath41 because of ( [ eq16 ] ) , the equations ( [ eq17 ] ) and ( [ eq18 ] ) are equivalent to our old equations ( [ eq10 ] ) and ( [ eq11 ] ) .
however , if we add ( [ eq17 ] ) and ( [ eq18 ] ) , we get @xmath42 and comparing with ( [ eq2 ] ) , we see that @xmath43 which is just the equation ( [ eq14 ] ) .
at last , let us consider the equation of motion for the fold @xmath44 of the rope ( figure [ fig2 ] ) .
its center of mass remains at rest .
therefore @xmath45 for the fold .
the mass @xmath28 of the fold remains also unchanged .
however we have non - zero momentum flux into the fold .
the pieces of the rope which join the fold in unite time bring the negative momentum @xmath46 with them , while the parts that disappear from the fold in unite time have in total the positive momentum @xmath47 .
therefore the total momentum flux into the fold is negative and equals to @xmath48 .
hence the equation of motion of the fold is @xmath49 which gives again the equation ( [ eq14 ] ) in the limit @xmath50 ( in fact , rope tensions differ slightly at the left and right ends of the bend - hence the term @xmath51 in the above equation .
however @xmath52 when @xmath50 . for finite @xmath24 , the calculation of @xmath51 is described in @xcite ) .
let us emphasize that ( [ eq15 ] ) is the correct equation of motion for systems of _ variable composition _
, the variable mass systems being just a particular case of such systems .
our last example above , the equation of motion of the rope s fold , shows clearly that the momentum flux may matter even in the case of constant mass systems , if their composition varies over time .
as was already mentioned , ( [ eq8 ] ) indicates that the weight of the sliding rope vanishes when the coordinate @xmath2 of the rope s right end becomes @xmath53 at that instant the velocity of the rope that follows from ( [ eq6 ] ) is @xmath54 what happens next actually depends on the arrangement how the rope is fixed on the peg @xcite .
if it goes through a frictionless channel in the peg , like a small duct , the negative weight which follows from ( [ eq8 ] ) will become a reality .
however , if the rope is not fixed in this manner on the peg , so that the arrangement is more like a pulley than a duct , the rope will start to whiplash after its velocity reaches ( [ eq19 ] ) .
assuming an ideal perfectly flexible rope , the motion will remain effectively one - dimensional nevertheless and can be analytically described @xcite .
moreover , this can be performed without using variable mass systems , after the rope s tension at the bend and the speeds of the ends of the rope are related by an educative guess @xcite .
the concept of momentum flux , however , allows to explain this relation for the tension more naturally .
figure [ fig3 ] shows the rope after the bend in the rope begins to move away from the peg .
the coordinates of the right and left ends of the rope are @xmath55 and @xmath56 respectively , while the coordinate of the ( infinitesimally small ) bend itself is @xmath2 ( note that @xmath57 ) .
as the rope is inextensible , we should have @xmath58 and , therefore , the bend moves with velocity @xmath59 in the non - inertial frame @xmath60 , where the bend is at rest , the right arm of the rope moves downward with velocity @xmath61 while the left arm moves upward with velocity @xmath62 therefore the situation in this system is analogous to the one considered in the previous section and hence we have momentum flux @xmath48 into the fold , where now @xmath63 in the limit @xmath50 , the equation of motion of the fold becomes ( note that the inertial force @xmath64 also goes to zero in this limit , along with @xmath65 , @xmath66 and @xmath51 ) @xmath67 which indicates that @xmath68 in the system @xmath60 , equations of motion for the right and left arms of the rope take the form @xmath69= \frac{m}{2l}(z_1-z)g - t+\pi-\frac{m}{2l}(z_1-z)\ddot{z } \label{eq25}\ ] ] and @xmath70= \frac{m}{2l}(z_2-z)g - t+\pi-\frac{m}{2l}(z_2-z)\ddot{z}. \label{eq26}\ ] ] the last terms in ( [ eq25 ] ) and ( [ eq26 ] ) represent inertial forces . because of ( [ eq23 ] ) , this equations are equivalent to simpler equations @xmath71 and @xmath72 after this point , the solution is essentially given in @xcite and goes as follows .
let us introduce the relative coordinate @xmath73 and note that the center - of - mass position is given by @xmath74=\frac{z_1+z_2}{2}+ \frac{z_m^2}{8l}-\frac{l}{2 } , \label{eq30}\ ] ] where we have used @xmath75 from ( [ eq20 ] ) .
equations ( [ eq29 ] ) and ( [ eq30 ] ) enable to express @xmath55 and @xmath56 through @xmath76 and @xmath77 : @xmath78 after the rope leaves the peg , the center of mass experiences a free fall with acceleration @xmath79 .
at the moment the whiplash begins , @xmath80 and we find from ( [ eq30 ] ) @xmath81 while ( [ eq6 ] ) indicates that @xmath82 and hence @xmath83\left .
\right |_{t=0}= \frac{1}{2}\sqrt{gl}.\ ] ] therefore , the center - of - mass motion is given by @xmath84 where the time @xmath85 is measured from the instant the rope leaves the peg . for the relative coordinate @xmath76
, we get from ( [ eq27 ] ) and ( [ eq28 ] ) @xmath86= \frac{1}{4}\,\frac{\dot{z}_m^2z_m}{(z_1-z)(z_2-z)}.\ ] ] but @xmath87 \left [ -\frac{1}{2}(z_1-z_2)+l\right ] = l^2-\frac{1}{4}\,z_m^2.\ ] ] therefore , finally @xmath88 this equation for @xmath76 seems formidable but , in fact , it is not as terrible as it looks . from ( [ eq33 ] )
we have @xmath89=0.\ ] ] therefore , @xmath90 if now we introduce a new variable @xmath91 through @xmath92 equation ( [ eq34 ] ) will take the form @xmath93 this equation can be easily integrated ( note that @xmath94 ) with the result @xmath95 this equation implicitly determines the function @xmath96 and hence , through ( [ eq35 ] ) , the function @xmath97 . relations ( [ eq31 ] ) , together with ( [ eq32 ] ) , then completely determine the motion of the rope up to the time @xmath98 when the rope becomes straight and the fold disappears ( note that at this moment @xmath99 and , therefore , @xmath100 ) .
our considerations of sliding rope above can shed light also to another interesting problem , the problem of falling chain @xcite .
one end of a folded flexible chain is fixed on a rigid support .
the another end is released in the manner of a bungee fall .
if initially the ends were placed close together , the chain motion is effectively one - dimensional and conceptually much like to the previous case of sliding rope . whether the falling part of the chain is in a free fall and moves with the acceleration @xmath79 , surprisingly was a subject of controversy for a long time @xcite .
it seems the overall consensus was in favor of the free fall .
sommerfeld gives a convincing argument @xcite ( albeit for an another falling chain problem , the steady link - by - link fall of a stationary chain from a resting heap ) , that the energy is not conserved in such problems due to inelastic collisions when new links are brought in motion .
therefore , one can expect that the falling arm is freely falling with acceleration @xmath79 and is gradually brought to rest link by link by inelastic impacts at the chain fold .
however , the last word in physics is an experiment , and the experiments @xcite , including the ones modeling bungee jumping @xcite , indicate that the tip of the falling chain moves with acceleration grater than @xmath79 .
what a surprise , this means that there is a downward pull at the fold even when the bungee rope is still slack !
`` this result is contrary to the usual experience with free falling objects and therefore hard to believe for many a person , even an experienced physicist '' @xcite .
however , the concept of momentum flux allows to explain the presence of this tension at the fold as naturally as it was in the case of sliding rope .
let the coordinates of the falling tip of the rope and of the fold are @xmath2 and @xmath101 respectively ( figure [ fig4 ] ) .
then @xmath102 , which gives @xmath103 in the non - inertial frame @xmath60 , where the fold is at rest , the left arm of the rope has the downward velocity @xmath104 while the right arm moves upward and its velocity is @xmath105 therefore , in the frame @xmath60 , we have a momentum flux @xmath106 in the fold with @xmath107 in its rest frame , equation of motion of the fold looks like @xmath108 and in the limit of the vanishing radius of the fold this equation reduces to @xmath109 we could stop here , as ( [ eq41 ] ) is just the mysterious downward pull in the bungee fall @xcite
. however , for completeness reasons , we proceed and re - derive some further results for falling chain .
equations of motion of the left and right subchains are ( in the non - inertial system @xmath60 and in the limit of vanishing radius of the fold ) @xmath110= \frac{m}{2l}(2l - z_f)g+t-\pi-\frac{m}{2l}(2l - z_f)\ddot{z}_f , \nonumber \\ & & \frac{d}{dt}\left [ \frac{m}{2l}z_f(0-\dot{z}_f)\right]= \frac{m}{2l}z_fg+t - f-\pi-\frac{m}{2l}z_f\ddot{z}_f , \label{eq42}\end{aligned}\ ] ] where @xmath5 stands for the weight of the falling chain .
equations ( [ eq42 ] ) simplify to @xmath111 and @xmath112 because of ( [ eq41 ] ) , equation ( [ eq43 ] ) takes the form @xmath113 again , this non - linear differential equation looks hard to solve , but note the following relation which follows from it : @xmath114= 2\dot{z}\left [ \left ( l-\frac{z}{2}\right ) \ddot{z}-\frac{1}{4}\dot{z}^2 \right ] = \ ] ] @xmath115.\ ] ] therefore , ( [ eq45 ] ) has the following first integral ( note the initial condition @xmath116 , when @xmath117 ) @xmath118 in fact , ( [ eq46 ] ) represents energy conservation @xcite . therefore , the velocity of the falling subchain is given by @xmath119 then from ( [ eq45 ] ) we get its greater than @xmath79 acceleration @xmath120 , \label{eq48}\ ] ] while ( [ eq44 ] ) gives the weight of the falling chain @xcite @xmath121 these results for the acceleration and weight can be generalized to include the mass of the subject attached to the falling tip of the chain ( or rope ) @xcite .
the finite mass of the bungee jumper regularizes ( [ eq47 ] ) , ( [ eq48 ] ) and ( [ eq49 ] ) , removing unphysical infinities in the velocity , acceleration and weight when @xmath122 .
the reader can easily modify our formulas above and reproduce these more general results .
we think , sliding rope problem is a good starting point to introduce the basics of variable mass mechanics to beginning students .
the problem looks simple but at closer inspection reveals some non - trivial features .
therefore , to disentangle it , we hope , will be both interesting and informative enterprise for students .
the machinery of variable mass mechanics , learned through this problem , can be applied to other similar situations .
for example , students can analyze falling chains and ropes and discover a surprising fact that bungee jumper falls with acceleration greater than @xmath79 , even though bungee rope is still slack in the first phase of bungee jumping . that a slack bungee rope can cause a downward pull may be surprising for students and even for experienced physicist @xcite .
however , to prove the reality of this pull , one should have not necessarily to go through a thrilling experience of real bungee jumping .
einstein s equivalence principle suffices to design a simple , elegant and convincing demonstration @xcite .
place a folded chain , or rope , on a smooth table and quickly pull one end of the rope .
you will find that the other end begins to move toward the bend , opposite to the direction of the pull .
therefore , there must exist a tension in the rope at the fold that drags the free end toward the bend .
however , in the non - inertial frame , where the yanked end is at rest , we have an effective gravity and the situation is just that is involved in the falling chain problem .
this demonstration also indicates that similar physics is involved in whipcracking .
however , in latter case the size of the fold is not small and the motion is no longer effectively one - dimensional . therefore , more advanced methods are needed to investigate this fascinating phenomenon @xcite , possibly discovered by dinosaurs millions of years ago @xcite .
it is surprising that without experiments our intuition can go astray even in relatively simple situations , like the falling chain problem .
first experiments @xcite , convincingly demonstrating a counter - intuitive result that the tip of the falling chain falls with acceleration greater than @xmath79 , have not had attracted any significant attention and this remarkable fact was largely ignored and unknown until new experiments @xcite have confirmed it .
modern physics operates with space , time and matter at the planck scale that can not be directly probed by experiment in the foreseeable future .
therefore , one faces the question `` can there be physics without experiments ? ''
the history of falling chain suggest that the answer is firmly `` no , there can not be any physics without experiments '' . without experiments we are always in a danger of being deceived by our metaphysical misconceptions .
for example , although variable mass systems were under investigation for more than two hundred years , unexpected and unexplainable anomalous behavior was observed , revealing flaws in current understanding of the dynamics of variable mass systems , when a new class of solid rocket motors was used in early 1980 s to power upper stages of several space missions @xcite .
as the last remark , note that the falling chain history can serve as an illustration of subtle interplay between theoretical and experimental methods in physics .
there were two theoretical paradigms in falling chain problem : energy conserving and the one that assumes maximal kinetic - energy loss due to completely inelastic collisions .
it may seem at first sight that the first paradigm was vindicated by experiment and the dispute is over . however , in any theoretical construction one deals with some idealized model of reality , real ropes and chains being somewhere in between of those two extreme idealizations .
an ideal , completely flexible , infinitesimally thin and unstretchable chain can be modeled as a series of point masses connected by massless strings . the behavior of such ideal folded chain , however , depends on the relative size of two specific length scales implicit in the model @xcite .
if the spacing between the point masses is large compared with the horizontal size of the fold ( figure [ fig5](a ) ) , each of the masses stops abruptly in an inelastic collision when it reaches the singularity ( kink ) in the rope .
the tension is not continuous along the kink , so that @xmath51 in ( [ eq40 ] ) does not vanish for infinitesimal bend and our previous analysis breaks at this point . on the contrary , if the spacing between the point masses is small compared with the horizontal span of the bend ( figure [ fig5](b ) ) , each mass comes to rest gradually and the system is energy conserving .
experiments @xcite indicate that the real ropes and chains involved in these experiments were more close to the energy conserving variant of figure [ fig5](b ) .
however , they say nothing about a possibility of experimental realization of another variant of figure [ fig5](a ) . moreover , this variant is more interesting , as an example of singular one - dimensional system with a kink
. therefore , it will be very illuminating to demonstrate that this type of the extreme limiting case of flexible chain could be also realized experimentally .
the author s attention to the problem of falling rope s weight was drawn by d. a. medvedev .
10 morin , d. , _ introduction to classical mechanics : with problems and solutions _ ( cambridge university press , 2008 ) .
sommerfeld , a. , _ mechanics _ ( academic press , new york , 1950 ) .
belchenko , yu .
i. , gilev , e. a. and silagadze , z. k. , _ problems in mechanics of particles and bodies _
( rcd , moscow - izhevsk , 2008 ) ( in russian ) . wong , c. w. and yasui , k. , _ falling chains _ , am . j. phys . * 74 * , 490 - 496 ( 2006 ) .
wong , c. w. , youn , s. h. and yasui , k. , _ the falling chain of hopkins , tait , steele and cayley _ , eur .
j. phys .
* 28 * , 385 - 400 ( 2007 ) .
tiersten , m. s. , _ force , momentum change , and motion _ , am . j. phys .
* 37 * , 82 - 87 ( 1969 ) .
siegel , s. , _ more about variable mass systems _ , am .
j. phys .
* 40 * , 183 - 185 ( 1972 ) .
galkin , m. g. , _ the dynamics of a falling chain : ii _ , am .
j. phys . * 57 * , 157 - 159 ( 1989 ) .
prato , d. and gleiser , r. j. , _ another look at the uniform rope sliding over the edge of a smooth table _ , am . j. phys . * 50 * , 536 - 539 ( 1982 ) .
galkin , m. g. and march , r. h. , _ the dynamics of a falling chain : i _ , am .
j. phys . * 57 * , 154 - 157 ( 1989 ) .
schagerl , m. , steindl , a. , steiner , w. and troger , h. , _ on the paradox of the free falling folded chain _ , acta mech .
* 125 * , 155 - 168 ( 1997 ) .
tomaszewski , w. , pieraski , p. and geminard , j .- c .
, _ the motion of a freely falling chain tip _ , am .
j. phys .
* 74 * , 776 - 783 ( 2006 ) .
kagan , d. and kott , a. , _ the greater - than - g acceleration of a bungee jumper _ ,
teacher * 34 * , 368 - 373 ( 1996 ) .
heck , a. , uylings , p. and kdzierska , e. , _ understanding the physics of bungee jumping _ , phys .
* 45 * , 63 - 72 ( 2010 ) .
wong , c. w. and yasui , k. , _ two falling - chain demonstrations based on einstein s equivalence principle _ , arxiv : physics/0609219v1 .
mcmillen , t , and goriely , a. , _ whip waves _
, physica d * 184 * , 192 - 225 ( 2003 ) .
mcmillen , t , and goriely , a. , _ shape of a cracking whip _
lett . * 88 * , 244301 ( 2002 ) .
krehl , p. , engemann , s. and schwenkel , d. , _ the puzzle of whip cracking uncovered by a correlation of whip - tip kinematics with shock wave emission _ , shock waves * 8 * , 1 - 9 ( 1998 ) .
myrhvold , n. p. and curie , p. j. , _ supersonic sauropods ?
tail dynamics in the diplodocids _ , paleobiology * 23 * 393 - 409 ( 1997 ) .
heywood , w. a. , hurwitz , h. , jr . and ryan , d. z. , _ whip effect in a falling chain _ , am .
j. phys .
* 23 * , 279 - 280 ( 1955 ) .
t hooft , g. , _ can there be physics without experiments ? challenges and pitfalls _ , int .
j. mod .
a * 16 * , 2895 - 2908 ( 2001 ) .
eke , f. o. and mao , t. c. , _ on the dynamics of variable mass systems _ , int .
j. mech .
30(2 ) * , 123 - 137 ( 2002 ) . | a simple mechanical problem is considered which we believe will help students to familiarize some concepts of mechanics of variable mass systems .
meanwhile they can even learn some thrilling physics of bungee jumping . |
in recent years higher dimensional gravity is investigated in various contexts such as string and brane - world models , and the gauge - gravity correspondence conjecture .
black holes solutions are of particular interest encoding information about the fundamental properties of the gravitational theory , and eventually making possible its experimental verification
. typical feature of higher dimensional gravity is that it allows greater abundance of black hole solutions than in four dimensions with qualitatively different properties @xcite .
ones of the most physically justified are kaluza - klein solutions , in which the extra dimension is compactified to a circle with a radius too small to be generally observed . in this paper
we will investigate in further details the black hole solutions in the five dimensional ( 5d ) einstein - maxwell - dilation gravity with kaluza - klein asymptotic .
we construct a new exact solution which represents a static black hole situated into external magnetic field , i.e. magnetized black hole .
the solution is only locally asymptotically kaluza - klein @xmath0 , but globally its spacelike cross - sections at infinity represent a hopf - fibration .
such geometry is characteristic for the euclidean taub - nut instanton , and in fact our solution reduces to a taub - nut instanton trivially embedded in 5d lorentzian spacetime in the limit when the event horizon vanishes .
for that reason the solution is also called a black hole on the taub - nut instanton being a part of a larger class of 5d black hole solutions on gravitational instantons @xcite .
we obtain the physical characteristics of the solution and derive the smarr - like relations . the calculations are closely related to our recent results on the thermodynamics of 5d black holes on asymptotically locally flat gravitational instantons @xcite , @xcite .
in addition , we investigate the hawking radiation of the solution by obtaining the greybody factor and the luminosity for emission of massless scalar particles in the low - energy regime .
similar calculations are already performed for 5d vacuum squashed black holes @xcite , @xcite , and they reveal that the emission spectrum deviates qualitatively from the case of a four - dimensional black hole .
thus , the hawking radiation of such black holes could be possibly used for experimental detection of extra dimensions .
we further show that , in distinction to the magnetized black holes with flat asymptotic , the external magnetic field influences the thermodynamical characteristics and the hawking radiation of the solution , and consequently can be used to tune the black hole parameters .
we consider the einstein - maxwell - dilaton gravity ( emd ) in five - dimensional spacetime with the action @xmath1 which leads to the field equations @xmath2 , \\
\nabla_{\mu}\nabla^{\mu}\varphi & = & -{a\over 2 } e^{-2a\varphi } f_{\nu\rho}f^{\nu\rho } , \nonumber \\ &
\nabla_{\mu}&\left[e^{-2a\varphi } f^{\mu\nu } \right ] = 0 , \nonumber\end{aligned}\ ] ] where @xmath3 is the ricci tensor for the spacetime metric @xmath4 , @xmath5 is the maxwell tensor , @xmath6 is the dilaton field and @xmath7 is the dilaton coupling parameter .
we are interested in solutions admitting three commuting killing vectors , one asymptotically timelike killing vector @xmath8 , and two spacelike killing vectors @xmath9 and @xmath10 , or more precisely , solutions with a group of symmetry @xmath11 .
furthermore , we require that the dilaton coupling parameter takes the particular value @xmath12 . under these conditions
we obtain the following exact solution @xmath13 \nonumber \\ & + & v^{-1/3 } \frac{r}{r+ r_{0}}\left(d\psi + r_\infty\cos\theta d\phi\right)^2,\end{aligned}\ ] ] where the metric function @xmath14 is given by @xmath15 the maxwell 2-form is expressed as @xmath16 with magnetic potentials @xmath17 the coordinates vary in the limits @xmath18 , @xmath19 and @xmath20 is a periodic coordinate with period @xmath21 .
the real parameter @xmath22 characterizes the electromagnetic field , @xmath23 is a scale parameter and @xmath24 denotes the position of a black hole horizon . the parameter @xmath25 is expressed as @xmath26 .
the solution contains coordinate singularity ( the so called misner string ) at @xmath27 ( or @xmath28 ) , resulting from the fact that no global chart exists for the solution manifold .
an atlas can be defined , consisting of several charts , if the coordinate @xmath29 is identified with a period @xmath30 , which ensures that the transition between the charts is well - defined .
consequently , the spatial cross - sections at @xmath31 are diffeomorphic to a hopf fibration of @xmath32 , and the solution is described as locally asymptotically flat according to the classification of @xcite .
the solution we obtained is a generalization of the vacuum squashed black hole @xcite , which is reproduced in the limit @xmath33 .
it also contains a new magnetized version of the gross - perry - sorkin monopole @xcite,@xcite
@xmath34 + \nonumber \\ & & v^{-{1\over3}}\frac{r}{r + r_0}\left(d\psi + r_\infty \cos\theta d\phi \right)^2 , \nonumber \\\end{aligned}\ ] ] as a limit at @xmath35 , when the black hole horizon vanishes .
the solution is characterized by two conserved gravitational charges - the mass and the tension @xcite , @xcite , which can be calculated by generalized komar integrals @xcite @xmath36 , \\ { \cal{t } } & = & - { 1\over 16\pi } \int_{s^{2}_{\infty } } \left[i_k \star d\xi - 2i_\xi \star d k \right ] , \nonumber\end{aligned}\ ] ] where @xmath37 is the killing field associated with time translations , @xmath38 is the killing field corresponding to the compact dimension , @xmath39 is the length of the @xmath40 fibre and @xmath41 is the base space of @xmath40-fibration at infinity . by direct calculation we obtain the result @xmath42 in addition to the adm mass , an intrinsic mass of the black hole is associated with the solution @xmath43 which can be expressed also in terms of the horizon area @xmath44 and surface gravity on the black hole horizon @xmath45 @xmath46 the surface gravity and the horizon area are determined by @xmath47 obviously satisfying ( @xmath48 ) .
as already discussed , the spacial cross - sections of the solution at @xmath31 represent a nontrivial @xmath40-fibration over @xmath49 .
consequently , the solution is characterized by a nut charge , which is proportional to the first chern class of the fibre bundle .
the nut charge can be calculated by the following komar - like integral @xcite @xmath50 where @xmath10 is the killing 1-form associated with the @xmath40 fibre at infinity , @xmath51 is its norm and @xmath52 is a two - dimensional surface , encompassing the nut . in our case
this is equivalent to the relation @xmath53 which was derived in @xcite for black holes on asymptotically locally flat gravitational instantons .
[ [ section ] ] in addition to the nut charge the solution is characterized by a related quantity , called a nut potential .
if we consider the 1-form @xmath54 where we use the dual electromagnetic potential @xmath55 satisfying @xmath56 , it can be proved that it is exact on the factor space @xmath57 @xcite .
consequently , there exists a globally defined potential @xmath58 , such as @xmath59 this relation determines the nut potential for the solution we investigate . the nut potential and the electromagnetic potential @xmath55 possess the following explicit form @xmath60 where they are normalized in such a way that the nut potential vanishes at infinity , and the electromagnetic potential vanishes on the black hole horizon .
the solution which we obtained does nt possess an electric or magnetic charge .
the electromagnetic field is characterized by the magnetic flux @xmath61 through the base space @xmath41 of the @xmath40fibration at infinity @xmath62 it is worth noting that the physical characteristics of the solution are affected by the external magnetic field , while such influence is not observed in the case of asymptotically flat magnetized black hole solutions @xcite , @xcite .
the dependence is realized through the parameter @xmath22 , which can be also expressed by means of the magnetic flux as @xmath63.\ ] ]
the physical characteristics of the solution , which we discussed in the previous section are connected by a set of geometrical relations , called smarr - like relations for the mass and the tension .
they can be derived by considering the komar integrals ( [ mt ] ) , reducing them to the factor space of the spacetime with respect to the isometry group @xmath64 , and performing a series of transformations by using the spacetime symmetries and the field equations .
the major part of the calculation coincides with the investigation of a related solution in @xcite , so we will mention here only the basic steps referring the reader to the corresponding article for further details .
the so - called interval structure associated with the solution @xcite , which determines the boundary of the factor space @xmath64 is of basic importance for the calculation . in our case
it consists of the following intervals * a semi - infinite space - like interval located at @xmath65 with direction @xmath66 ; * a finite timelike interval located at @xmath67 with direction @xmath68 corresponding to the black hole horizon ; * a semi - infinite space - like interval at @xmath69 with direction @xmath70 , where the directions of the intervals are determined with respect to a basis of killing vectors @xmath71 . as a starting point of the calculation we consider the komar integral for the tension and reduce it to the factor space @xmath72\end{aligned}\ ] ] where the integration is now performed over the semicircle representing the boundary of the two - dimensional factor space at infinity . using the stokes theorem the integral can be further expanded into a bulk term over @xmath64 and an integral over the rest of the boundary of the factor space represented by the interval structure @xmath73 @xmath74 - { l\over 8}\sum_i \int_{i_i } \left[i_\eta i_k \star d\xi - 2i_\eta i_\xi \star d k \right ] \nonumber \\ & = & \frac{1}{2}m_h + { l\over 4 } \int_{i_l\bigcup i_r } i_\eta i_\xi \star d k + { l\over 8 } \int_{\hat{m } } \left[d i_\eta i_k \star d\xi - 2d i_\eta i_\xi \star d k \right],\end{aligned}\ ] ] where @xmath75 is the intrinsic mass of the black hole ( [ mh ] ) . the bulk integral can be expressed as @xmath76 & = & { l\over 4 } \int_{\hat{m } } \left[i_\eta i_k \star r(\xi ) - 2i_\eta i_\xi \star r(k ) \right ] \nonumber \\ & = & - { l\over 2 } \int_{\hat{m}}\left [ d a_\psi
\wedge i_\eta i_\xi e^{-2a\varphi}\star f + d a_\phi \wedge i_k i_\xi e^{-2a\varphi}\star f\right ] , \nonumber\end{aligned}\ ] ] taking account of the ricci - identity @xmath77 and deriving from the field equations the relation @xmath78 both of which apply for any killing field @xmath10 . we can further simplify the expression using the stokes theorem @xmath79= \nonumber \\ & & - { l\over 2 } \int_{arc(\infty ) } a_\psi i_\eta i_\xi e^{-2a\varphi}\star f + a_\phi i_k i_\xi e^{-2a\varphi}\star f - \nonumber \\ & & { l\over 2 } \int_{i_l\bigcup i_r}\left [ a_\psi i_\eta i_\xi e^{-2a\varphi}\star f + a_\phi i_k i_\xi e^{-2a\varphi}\star f\right ] , \nonumber\end{aligned}\ ] ] and showing that the integral over the boundary of the two - dimensional factor space at infinity vanishes due to the explicit form of the electromagnetic field .
thus , we obtain the relation @xmath80 , \nonumber\end{aligned}\ ] ] which can be also represented as @xmath81 where @xmath58 is the value of the nut potential on the horizon .
we introduce the potentials @xmath82 and @xmath83 .
they are constant of the left and right semi - infinite intervals respectively , and satisfy @xmath84 . in this way we obtain the smarr - like relation for the tension @xmath85 where @xmath86 is the asymptotic value of the electromagnetic potential ( @xmath87 ) . the relation can be expressed also by means of the magnetic flux @xmath61 at infinity ( @xmath88 ) @xmath89 where we have introduced an effective current sourcing the magnetic field @xmath90 in a similar way if we take advantage of the komar integral definition of the adm mass we can derive the smarr - like relation for the mass @xmath91
it is well known that in a semi - classical approximation black holes emit particles as a black body with temperature proportional to the surface gravity of the horizon @xcite .
the emission is characterized by an exact thermal spectrum only at the vicinity of the horizon .
if observed at infinity , it is modified by a factor , called greybody factor , due to the interaction of the radiation with the non - trivial geometry of the curved spacetime acting as an effective potential barrier .
the greybody factor reflects the probability of an out - going wave with a certain frequency to reach infinity , and it is equivalent to the absorption probability for the same frequency mode , i.e. the probability for an in - coming wave to be absorbed by the black hole
. there exists a simple procedure for calculating the absorption probability in cases when the differential equation describing the propagation of a field in a particular spacetime is separable .
then it can be reduced to a schrdinger - type equation for some radial coordinate with an effective potential encoding the properties of the propagating field and the spacetime geometry , and the problem reduces to the quantum mechanical problem of wave scattering by a potential barrier .
although the schrdinger - type equation can not be solved exactly in most of the cases , an approximation scheme can be applied in the low - energy limit , when the wave length of the emitted particles is larger than the characteristic scale of the black hole horizon .
classical examples of such calculations are @xcite-@xcite . in this section
we will calculate the absorption probability of the magnetized black hole solution we presented in section 2 for massless scalar fields .
we consider the corresponding field equation in the spacetime described by ( @xmath92 ) @xmath93 the equation is separable and possesses the following type of solutions @xmath94 with radial and angular functions @xmath95 and @xmath96 satisfying a set of differential equations @xmath97 where we have introduced the functions @xmath98 , @xmath99 , and @xmath100 for convenience . the angular equation is solved in the so called spin - weighted spherical harmonics @xmath101 with parameters @xmath102 and @xmath103 .
the radial equation can be transformed into a schrdinger - like equation @xmath104 by making the coordinate transformation @xmath105 and introducing a new radial function @xmath106 such as @xmath107 .
the potential @xmath108 is given in terms of the initial radial coordinate by the following expression @xmath109 ( 0,0)(0,0 ) ( -25,102 ) ( 120,-10)r we consider the scattering of an incident wave at infinity by the effective potential ( @xmath110 ) .
taking into account the properties of the spacetime , we require that no outgoing wave should be present at @xmath111 , as a black hole horizon is located there .
thus , the physically relevant solutions should obey the boundary conditions @xmath112 where @xmath113 and @xmath114 are the transmission and reflection coefficients , respectively . by definition the absorption probability
@xmath115 is proportional to the transmission coefficient , i.e. @xmath116 . the schrdinger - like equation can not be solved exactly , for that reason we will find the absorption probability only for low energies of the scalar field , when an approximation scheme can be applied .
the low energy regime implies frequencies that satisfy @xmath117 where @xmath118 is the horizon radius , and @xmath119 is the black hole temperature .
then we can define several regions in spacetime , solve in each of them an approximate form of the scrdinger - like equation which is valid in the particular region , and match the obtained solutions at the common region boundaries .
the local solutions can be matched with a high accuracy because in the low energy limit the wave - length of the scalar field is much larger than any of the characteristic scales associated with the black hole .
usually the following three regions are defined : a near horizon region , where @xmath120 , and the frequency is much larger that the effective potential , @xmath121 , an intermediate region , where the effective potential is much larger than the frequency , @xmath122 , and asymptotic region , where @xmath123 .
it is reasonable also to assume that @xmath124 as the leading contribution to the absorption probability in the low energy limit comes from this mode . applying the approximation scheme we briefly described , we obtain that the radial equation ( @xmath125 ) reduces to the following form @xmath126 in the near horizon region .
it can be solved by using the transformation @xmath127 , thus obtaining a solution @xmath128 where we have taken into account the boundary condition that only ingoing waves should be present at the horizon . in the limit @xmath129 , which corresponds to the boundary of the near - horizon region with the intermediate region ,
this solution becomes @xmath130.\end{aligned}\ ] ] next we consider the intermediate region with a characteristic scale @xmath131 .
we can neglect the frequency compared to the effective potential there and obtain the approximate equation @xmath132 which is solved by @xmath133 the solution has the following limit at @xmath134 , corresponding to the boundary with the near - horizon region @xmath135 which should be equivalent with the near - horizon solution ( @xmath136 ) , and another limit at @xmath137 @xmath138 corresponding to the boundary with the asymptotic region .
finally , in the asymptotic region @xmath139 the following approximate equation applies @xmath140 it can be solved analytically as @xmath141 and the solution tends to @xmath142 in the limit @xmath143 .
this limit overlaps with the intermediate region , since @xmath144 is satisfied for its characteristic scale @xmath145 when low energies are considered . consequently , the expression ( @xmath146 ) should be equivalent to the intermediate region solution ( @xmath147 ) . as a result of the described calculation
the following matching conditions should be imposed on the integration constants of the local solutions in the three regions @xmath148 reflecting the equivalence of solutions ( @xmath136 ) and ( @xmath149 ) , as well as ( @xmath147 ) and ( @xmath146 ) .
the solution in the asymptotic region can also be presented in the form @xmath150 hence , the reflection coefficient is given by @xmath151 , where the constants @xmath152 can be expressed by means of the frequency and the spacetime parameters by considering the matching conditions ( @xmath153 ) .
consequently , we obtain the absorption probability @xmath154 where @xmath44 is the horizon area ( @xmath155 ) and @xmath156 is the length of the compact dimension . the luminosity of the emission of a particular kind of uncharged particles with absorption probability @xmath157 is determined by the expression @xmath158 where @xmath159 is the black hole temperature .
taking advantage of eq .
( @xmath160 ) we obtain the luminosity of emission of massless scalar particles at low energies for our solution @xmath161 these expressions reveal that the absorption probability and the luminosity are affected by the external magnetic field , with eq .
( @xmath162 ) giving the explicit dependence on the magnetic flux @xmath61 at infinity .
the 5d einstein - maxwell - dilation equations with a dilaton coupling parameter @xmath12 possess the following more general black hole solution @xmath163 the 1-form @xmath164 and the metric functions @xmath165 and @xmath166 are given by @xmath167 and the maxwell 2-form is expressed as @xmath168 with electromagnetic potentials @xmath169 \cos\theta , \\
\nonumber\\ & & a_{\psi } = \frac{1}{2}\cosh\gamma\sinh\gamma \frac{r^2_{\infty}}{r+ r_{0}}.\end{aligned}\ ] ] the coordinates vary in the limits @xmath18 , @xmath19 and @xmath20 is a periodic coordinate with period @xmath21 .
the parameters @xmath22 , @xmath170 and @xmath118 take the same ranges as in section 2 , @xmath171 is expressed as @xmath172 , and @xmath173 is a second real parameter characterizing the electromagnetic field . in the limit @xmath174 the solution reduces to the magnetized black hole which we discussed in section 2 .
the solution possesses coordinate singularities at @xmath27 ( or @xmath28 ) , which can be resolved if the coordinates @xmath175 and @xmath29 are identified with periods @xmath176 and @xmath177 .
consequently , the cross - sections at @xmath178 , and at @xmath179 represent hopf fibration of @xmath32 with @xmath40 fibres parameterized by the @xmath29 and @xmath175 coordinates , respectively . due to the periodic identification of the time
coordinate the solution contains closed time - like curves , meaning that the causality of the spacetime is violated . in the limit at @xmath174 , however , the fibre bundle associated with the time coordinate becomes trivial and no closed timelike curves occur .
p. n. would like to thank daad for the support , and the oldenburg university for its kind hospitality .
the partial financial support by the bulgarian national science fund under grant dmu-03/6 , and by sofia university research fund under grant 148/2012 are also gratefully acknowledged .
r. emparan , h. reall , living rev .
* 11 * ( 2008 ) 6 [ arxiv:0801.3471 [ hep - th ] ]
. y. chen , e. teo , nucl .
b * 850 * ( 2011 ) 253 [ arxiv:1011.6464 [ gr - qc ] ] .
p. nedkova , s. yazadjiev , phys .
d * 84 * ( 2011 ) 124040 arxiv:1109.2838 [ hep - th ] .
p. nedkova , s. yazadjiev , phys .
d85 ( 2012 ) 064021 [ arxiv:1112.3326 ] | we construct a new exact solution in the 5d einstein - maxwell - dilaton gravity describing a magnetized squashed black hole .
we calculate its physical characteristics , and derive the smarr - like relations . the hawking radiation of the solution
is also investigated by calculating the greybody factor for the emission of massless scalar particles in the low - energy regime . in distinction to the case of asymptotically flat magnetized black holes , the thermodynamical characteristics and the hawking radiation of the solution depend on the external magnetic field , and consequently can be tuned by its variation . |
the investigations conducted prior to the launches of chandra and xmm have speculated that a significant fraction of the x - ray background ( xrb ) is produced by discrete sources , predominantly agns ( e.g. hasinger etal .
1998 ) . with its superb spatial resolution ,
chandra resolved nearly 100% of the x - ray background in the 0.58.0 kev regime into discrete sources ( e.g. mushotzsky etal .
2000 ) , and thus changed the theme of the xrb study into investigating the nature of the discrete x - ray sources and their constraints on galaxy evolution . with the active optical and infrared follow - ups around the two deepest chandra fields ,
chandra deep field north ( cdf - n ; brandt etal .
2001b ) and south ( cdf - s ; giacconi etal . 2002 ) , we have gained extensive knowledge about the faint x - ray source population .
it is now generally believed that the majority of these sources are dust - obscured type - ii agns at @xmath11=02 , while a smaller fraction of them are classic type - i agns out to @xmath12 ( e.g. , alexander etal .
2001 ; tozzi etal . 2001 ; norman etal . 2002 ; rosati etal .
2002 ) .
however , a significant number of these x - ray sources are still optically unidentified and thus their nature remains uncertain . within an [email protected] area at the center of the cdf - n , the multi - wavelength investigation of alexander etal .
( 2001 ) indicates that fifteen , or about 10% of the total x - ray sources do not have counterparts to a 2@xmath14 limit of @xmath3 = 25.3 mag ( all magnitudes quoted in this letter are in the vega system unless otherwise noted ) .
five out of these fifteen sources , however , have near - ir counterparts on the @xmath15-band images of barger etal . ( 1999 ) . similarly , giacconi etal .
( 2002 ) concluded that 59 , or about 17% of the total x - ray sources in the cdf - s were not optically identified at a limit of @xmath0 = 26.026.7 mag .
these facts raise a very natural question : are these optically invisible sources merely the fainter tail of the already identified population , or do they constitute a separate population of their own ? in this letter , we study the nature of the optically unidentified sources in giacconi etal .
( 2002 ) , which we shall refer to as `` @xmath0-unidentified sources '' .
we utilize the deep vlt / isaac @xmath2 and vlt / fors1 @xmath3-band data , both of which are now publicly available from the eso science archive .
the analysis of these data are described in 2 , and the ir and @xmath3-band identifications are shown in 3 , where we presnet ir identifications of six such x - ray sources . in
4 we explore the possible nature of these sources based on their optical to near - ir photometric properties .
we conclude with a summary in 5 .
as part of the supporting observations of the _ great observatories origins deep survey _ ( goods ; dickinson & giavalisco 2002 ) fields ,
the eso vlt / isaac has imaged a large portion of the cdf - s in @xmath16 , @xmath17 , and @xmath18-bands ( hereafter @xmath2 ) . the first release of these fully processed and stacked data consists of six continuous fields , and covered about 50 @xmath19 at the center of the cdf - s . to further constrain the optical characteristics of the chandra sources ,
we also independently reduced the whole set of cdf - s @xmath3-band vlt / fors1 images ( 3 hours of exposure , eso prog .
64.o-0621(a ) , pi gilmozzi ) retrieved from the eso science archive .
these @xmath3-band data , consisting of four slightly overlapped fields , extend @xmath20 150 @xmath19 and fully cover the area that the vlt / isaac has imaged in the @xmath2 .
the photometry on both the @xmath2 data and the @xmath3-band data was done by using the sextractor routine of bertin & arnouts ( 1996 ) . for each isaac field ,
the @xmath2 images were combined into a master stack , which was then used for extracting sources and defining apertures . while the photometry was performed on the individual passbands , a same aperture defined from the master stack was used for a given source in each passband ( i.e. , matched - aperture photometry " ) .
for source detection , we used a @xmath21 gaussian smoothing kernel with the fwhm of 2.5 pixels , which is approximately the same as the fwhm of a point source psf ( @xmath22 ) on the master stack .
we set 1.5 @xmath14 as the detecting threshold , and required at least 5 connected pixels above the threshold for a source to be detected .
as the vast majority of the sources are extended , we used total magnitude ( corresponding to the _ mag - auto _ option in sextractor ) for the photometry on the individual passbands .
the zeropoints recorded in the image headers were used to convert counts into vega magnitudes .
we assessed the @xmath2 limits in the following way . the representative error reported by sextractor
was used to calculate the s / n of each extracted source , assuming the reported magnitude error ( @xmath23 ) and the s / n follow the simple relation as @xmath24 . only the sources with @xmath25
were included in the assessment .
the source count histogram was constructed for each band , and the magnitudes at which the counts dropped to 50% of the peak values were defined as the survey limits .
these numbers are : @xmath16=24.2 , @xmath17=23.4 and @xmath18=23.2 mag .
source detection and photometry on the @xmath3-band stacks were done in the same manner as in analyzing the isaac data excepted that the detection threshold was lowered to @xmath26 to include as faint sources as possible . as the eso imaging survey ( eis ; arnouts etal .
2002 ) covered the entire cdf - s , we used their @xmath3-band source catalog to calibrated our photometry .
we first removed the galactic extinction ( @xmath27=0.02 mag ) that they applied to their results , and converted the magnitudes from ab system , which they used , to vega system by using @xmath28 .
the survey limits , derived in the same way as in the isaac data analysis , are approximately 25.1 25.2 mag accounting for field - to - field variation .
out of the 346 chandra sources , 77 have their locations falling within the @xmath2 coverage , among which twelve are @xmath0-unidentified sources .
we found that six of these twelve objects could be matched with the @xmath2 sources within a matching radius of 1.5 " , all of which were strikingly significant detections ( at 3.59.0@xmath14 level ) in the @xmath18-band .
after visually examining these matched sources on screen and comparing against the @xmath0-band image stamps given in giacconi etal .
( 2002 ) , we concluded that these six ir counterparts were real identifications .
we then cross - matched the @xmath0-unidentified sources with our @xmath3-band images . while 35 of them have their locations within the @xmath3-band coverage , only one counterpart was found .
this fact further suggests that most of these objects are indeed optically faint .
however , the only counterpart , which is also one of the six @xmath2 sources , was detected at above @xmath29 level .
table 1 lists the positions and the photometric characteristics of the six objects that we identified , together with their x - ray hardness ratios ( @xmath30 ) taken from giacconi etal .
these six sources are of moderate x - ray fluxes in 0.52 kev , with two of them falling toward the faint end ( @xmath31 erg / s/@xmath32 ) .
we note that they tend to have negative x - ray hardness ratios .
in fact , three of them were detected in the soft channel only ( @xmath33 ) .
assuming that their underneath x - ray sources are agns , it is possible to determine their types based on their @xmath30 values . if we use @xmath34 as the dividing line between low x - ray luminosity ( @xmath35 ) type - ii agn and high @xmath35 type - i agn ( e.g. rosati etal .
2002 ) , three of these six chandra sources are type - i agns while the remaining two are type - ii agns .
however , such a division should be used with caution here , as our case is of small number statistics ( cf .
fig . 3 of rosati etal .
2002 ) . the excellent image quality ( point source fwhm @xmath36 in both ir and @xmath3-band ) allowed us to at least determine if these sources are of extended morphology .
based on the @xmath18-band image , where the s / n of the objects is the highest , we determined that three of them were extended sources , and two of them were point sources .
the remaining one is hard to tell because it is close to the @xmath18-band field edge and its background is much noisier than that of the others .
although these @xmath0-unidentified sources could simply be the fainter tail of the brighter objects that have been optically identified , we should not limit our investigation to only this possibility .
for this reason , we should consider a wide variety of candidates when exploring the nature of these x - ray source hosts .
as alexander etal .
( 2001 ) argued , it is very unlikely that the objects in the galaxy could be a major contribution to optically faint x - ray sources .
this possibility is even smaller for the cdf - s because it is at high galactic latitude .
thus we consider four types of extragalactic objects as the primary candidates , namely , elliptical galaxies ( e / s0 ) , agns / qsos ( agn ) , irregular galaxies ( irr ) , and very young galaxies at high redshift ( @xmath37 ; high - z ) .
we do not include spiral galaxies for the sake of simplicity , as their photometric properties in the @xmath3-band and the near - ir lie in between the ellipticals and the irregulars .
we compare the six ir counterparts against these four types of candidate ins the @xmath38 color space .
we consider the redshift range of @xmath39 for high redshift young galaxies , and @xmath8 for all the other three types of candidates .
we also consider the effect of internal dust extinction of @xmath40=0.13 mag , following the standard milky way extinction law .
for the wavelength range shortward of 2000@xmath41 , we use the average values of the lmc , smc and the galaxy ( e.g. calzetti , kinney & storchi - bergmann 1994 ) . because five of these six objects are only visible in the three ir passbands , it is not realistic to perform rigorous photometric redshift analysis based on such limited information . at this stage
, we chose to use color - color diagrams as our primary diagnostic tools , which are less quantitative but more robust .
the invisibility of these sources in the @xmath0-band and blueward was used as a consistency check for each of the four possibilities , and was also used as one of the inputs ( detailed below ) to constrain the rest - frame luminosity of the sources .
we simulated the colors of the candidate objects by using their representative spectral templates .
the templates for e / s0 and irr were taken from coleman , wu & weedman ( 1980 ) with the extension to uv and ir according to bruzual & charlot ( 1993 ) , and the one for agn was taken from vanden berk etal .
( 2002 ; using @xmath42 to extend its continuum to the near - ir ) . for young
, high redshift galaxies , we used a 0.1 gyr age , solar metalicity template from bruzual & charlot ( 1993 ) .
these rest - frame spectral templates were redshifted in @xmath43=0.5 step , and attenuated for the cosmic h i absorption according to madau ( 1994 ) .
the resulting spectra were then convolved with the system responses , mainly the filter transmission and the detector quantum efficiency , to generate the colors in the ab magnitude system . when considering dust extinction , the colors were adjusted according to the rest - frame spectral range sampled by the relevant passbands .
the colors were finally transfered into the vega system by adopting the appropriate zeropoints ( e.g. waddington etal .
1999 ) .
1 shows the @xmath44 vs. @xmath45 color - color diagrams in four separate panels for the four cases .
the field objects with @xmath46 in all three bands are shown as the green dots , and the chandra sources that have been identified in @xmath0-band are shown as the blue squares .
the six ir counterparts are plotted with different symbols based on their hardness ratios : the three solid triangles are the ones with @xmath33 , the two open hexagons that have @xmath47 , and the open square is the source that has @xmath48 .
their error bars indicate the photometric errors in their colors , which are calculated as the square - root of the quadratic sum of their photometric errors in the corresponding passbands .
the unreddened and reddened colors of the simulated objects at different redshifts are plotted in black crosses and solid red squares , respectively ( in the case of e / s0 , only unreddened colors are shown ) . for clarity ,
only the optimal results are plotted for the reddened colors .
these optimal reddening values are @xmath49 and 1.6 mag for irr , agn and high - z cases , respectively .
a couple of redshifts are marked to indicate the direction along which the color of the simulate objects changes with redshift . similarly , fig .
2 shows the @xmath44 vs. @xmath50 color - color diagrams . the high - z case is not included in this figure , because galaxies at @xmath51 5 will _ drop - out _ from the @xmath3-band and thus they can not put meaningful @xmath50 constraint on those ir counterparts that are also @xmath3-band drop - outs .
all the legends are the same as in fig.1 , except that the five of the six ir counterparts that were not detected in the @xmath3-band are not plotted with error - bars .
instead , we put upward arrows on them to indicate their @xmath50 color limits
. it is intriguing to note that the location of those optically visible chandra sources in the ir color space is bifurcated , as can be clearly seen in fig .
the upper branch coincides with the e / s0 track from @xmath4 while the lower branch seems to be consistent with the agn track from @xmath5 .
this interpretation does not conflict with the investigation on the brighter sources in rosati etal .
( 2002 ; their fig . 5 . ) .
the six objects that we identified lie on the upper branch and its extension following the e / s0 track up to @xmath52 ( see below ) . using the two color - color diagrams and the source brightness , we are able to set stringent constraints on the nature of these six ir counterparts .
we comment on the above - mentioned four possibilities individually .
@xmath53 from fig . 1 and 2 it is obvious that the six sources have colors consistent with unreddened e / s0 galaxies at @xmath11=0.52.5 , and introducing internal extinction does not improve the fit . considering the brightness of these objects , this interpretation is also consistent with their invisibility in the @xmath0-band and blueward . while no
. 593 ( the lowest triangle in fig .
1 ) , which was detected in the @xmath3-band as well , is probably at @xmath11=0.5 , the other five objects are likely at @xmath54 .
if we use @xmath11=2.0 as their representative redshift , their averaged rest - frame @xmath55-band absolute magnitude is @xmath56 = 20.1 mag ( in ab system ) , indicating that they are giant ellipticals .
in contrast , no .
593 has only @xmath57 mag if it is at @xmath11=0.5 , falling in the dwarf elliptical regime .
@xmath58 while they can reasonably explain the colors of the optically bright chandra sources , unreddened agns at @xmath8 can not produce colors similar to the six sources ( see especially fig .
if internal reddening is introduced , the matching will be improved .
however , this interpretation involves both the extinction and the redshift as free parameters , and requires both of them be adjusted for each source individually .
if the extinction is forced to be fixed , the best fit is obtained with @xmath59 , which allows three of the six source ( open symbols ) to be explained , i.e. , at @xmath60 01 or @xmath60 45 .
if we take the first interpretation for these @xmath61 objects , their absence from @xmath3-band and blueward means they can only be at @xmath60 0.81.0 to have a reasonable intrinsic luminosity that still qualified for agns ( @xmath62(38004200@xmath63 ) @xmath64 to @xmath65 mag after de - reddening ) .
if the later interpretation is real , on the other hand , their absolute magnitudes make them fall in the low - luminosity end of qsos ( @xmath62(38004200@xmath63 ) @xmath66 to @xmath67 mag after de - reddening ) .
the cumulative qso surface density inferred from this interpretation ( 23 in 50 @xmath19 ) is consistent with the qso luminosity function extrapolated from that is known at lower redshifts ( e.g. boyle etal .
particularly , we noted that no .
593 did not fit in this interpretation no matter how the extinction and the redshift were adjusted .
@xmath68 similar to the agn case , the colors of unreddened irregular galaxies match the optically bright chandra sources but can not match those of the six @xmath0-unidentified sources . introducing internal reddening can not solve the problem , as fig . 2 shows .
thus irregular galaxies are not likely a viable explanation to the nature of the six sources .
@xmath69 the colors of unreddened high - z galaxies are not consistent with the six sources .
fine - tunning internal reddening value can produce a resonable fit either at the redshift range @xmath70 or at @xmath71 .
the reddening value thus required is @xmath72 mag .
however , in either case , the cumulative surface density of such galaxies are at least one order of magnitude higher than a reasonable luminosity function would predict ( following the same methodology as in yan etal .
thus we conclude that this hypothesis is not a viable interpretation either .
we present deep optical and near - ir identifications of the @xmath0-unidentified x - ray sources in the cdf - s , using the deep vlt / isaac and fors1 imaging data publicly available from the eso science archive .
alexander etal .
( 2001 , 2002 ) did a similar work over a comparable size of area around the cdf - n , who largely focused on the optically faint _ but visible sources_. the vlt data that we used reached at least two magnitudes deeper than theirs in the near - ir , and was at least twice as deep in the @xmath3-band .
more importantly , the three near - ir bands used in this study allow us to further constrain the nature of such sources .
we found six ir counterparts for the cdf - s 1ms sources that were not optically identified to @xmath0=26.126.7 mag , making it possible to carry out spectroscopic study in the future .
the x - ray hardness ratio of these six chandra sources are more negative , opposite to the trend found among the similar , but optically brighter sources in the cdf - n ( alexander etal .
the multi - band photometric properties of these six sources are compared against those of the optically visible chandra sources in the same field .
we found that the location of these optically brighter chandra sources in the near - ir color space was bifurcated , with the color of one branch being consistent with that of e / s0 galaxies at @xmath4 and the other branch being consistent with that of unreddened agn / qsos at @xmath5 .
the six counterparts that we identified seem to lie on the e / s0 branch and its extension , suggesting that these x - ray source hosts are mostly luminous e / s0 galaxies ( @xmath6 mag in ab system ) at @xmath7 . such luminous ellipticals that host x - ray sources might be similar to those found by cowie etal .
( 2001 ) , which are amplified by a foreground cluster . on the other hand , some of these six sources can also be explained by agn / qsos over a wide redshift range ( @xmath8 ) if various internal extinction ( @xmath91 mag ) is allowed .
however , the later interpretation requires fine - tuning extinction together with redshift for these objects individually .
if they are indeed agn / qsos , the most luminous of them is just barely qualified for being a qso . finally , we point out that neither high - redshift ( @xmath10 ) star - forming galaxies nor irregular galaxies at lower redshift can be a viable explanation to the nature of these six counterparts .
alexander , d. , et al .
2001 , , 122 , 2156 , 2002 , , 123 , 1149 boyle , b. j. , et al .
2000 , , 317 , 1014 bertin , e. & arnouts , s. 1996 , , 117 , 393 brandt , w. , et al 2001a , , 122 , 1
, 2001b , , 122 , 2180 bruzual , a. & charlot , s. 1993 , , 405 , 538 calzetti , d. , kinney , a. & storchi - bergmann , t. 1994 , , 429 , 582 cowie , l. , et al .
2001 , , 551 , l9 coleman , g. , wu , c .- c . & weedman , d. 1980 , , 43 , 393 dickinson , m. & giavalisco , m. 2002 , astro - ph/0204213 giacconi , r. , et al .
2002 , , 139 , 369 hasinger , g. , et al .
1998 , , 329 , 482 madau , p. 1995
, , 441 , 18 norman , c. , et al .
2002 , , 571 , 218 rosati , p. , et al .
2002 , , 566 , 667 vanden berk , d. , et al .
2002 , , 122 , 549 waddington , i. , et al .
1999 , , 526 , l77 yan , h. , et al .
2002 , to be published in ( astro - ph/0208080 ) i d & ra & dec ( j2000 ) & @xmath3 & @xmath16 & @xmath17 & @xmath18 & p / e & hr + 79 & 3:32:38.04 -27:46:26.2 & @xmath73 & [email protected] & [email protected] & [email protected] & p & -0.42 + 593 & 3:32:14.79 -27:44:02.5 & [email protected] & [email protected] & [email protected] & [email protected] & e & -1.00 + 221 & 3:32:08.91 -27:44:24.8 & @xmath73 & [email protected] & [email protected] & [email protected] & p & -1.00 + 515 & 3:32:32.17 -27:46:51.5 & @xmath73 & [email protected] & [email protected] & [email protected] & e & 0.41 + 561 & 3:32:22.44 -27:45:43.8 & @xmath73 & [email protected] & [email protected] & [email protected] & e & -1.00 + 201 & 3:32:39.06 -27:44:39.3 & @xmath73 & [email protected] & [email protected] & [email protected] & ... & -0.06 + | the chandra deep field south ( cdf - s ) 1ms exposure produced a catalog of 346 x - ray sources , of which 59 were not visible on the vlt / fors1 and the eso - mpi / wfi deep @xmath0-band images to a limit of @xmath1=26.126.7 mag .
using the first release of the eso vlt / isaac @xmath2 data on the cdf - s , we identified six of the twelve such objects that were within the coverage of these ir observations .
the vlt / fors1 @xmath3-band data further confirms that five of these six objects are undetected in the optical .
the photometric properties of these six counterparts are compared against those of the optically brighter counterparts of chandra sources in the same field .
we found that the location of these optically brighter chandra sources in the near - ir color space was bifurcated , with the color of one branch being consistent with that of e / s0 galaxies at @xmath4 and the other branch being consistent with that of unreddened agn / qsos at @xmath5 .
the six counterparts that we identified seemed to lie on the e / s0 branch and its extension , suggesting that these x - ray source hosts are mostly luminous e / s0 galaxies ( @xmath6 mag in ab system ) at @xmath7 . on the other hand ,
some of them can also be explained by agn / qsos over a wide redshift range ( @xmath8 ) if a range of internal extinction ( @xmath91 mag ) is allowed .
however , the later interpretation requires fine - tuning extinction together with redshift for these objects individually .
if they are indeed agn / qsos , the most luminous of them is just barely qualified for being a qso .
finally , we point out that neither high - redshift ( @xmath10 ) star - forming galaxies nor irregular galaxies at lower redshift can be a viable explanation to the nature of these six counterparts . |
one of the most amazing predictions offered by the quantum theory of light is what has been called _ vacuum fluctuations _ : even in absence of photons ( vacuum ) the value of the fluctuations of some observables are different from zero .
these fluctuations can not be removed by improving the experimental instrumental , and hence , they are a source of non - technical noise ( _ quantum noise _ ) which seems to establish a limit for the precision of experiments involving light . during the late seventies and mid - eighties of the past century ways for overcoming this fundamental limit were predicted and experimentally demonstrated @xcite . in the case of the quadratures of light ( equivalent to the position and momentum of a harmonic oscillator ) , the trick was to eliminate ( _ squeeze _ ) quantum noise from one quadrature at the expense of increasing the noise of its canonically conjugated one , in order to preserve their heisenberg uncertainty relation .
states with this property are called _ squeezed states _ , and they can be generated by means of nonlinear optical processes .
in particular , the largest levels of squeezing are obtained by using nonlinear resonators operating near threshold , this level being degraded quickly as one moves away from that critical point . up to date , the best squeezing ever achieved is a 90% of noise reduction with respect to vacuum ( the so - called _ standard quantum limit _ ) @xcite , and _ degenerate optical parametric oscillators _ ( dopos ) have been used to this aim .
a dopo consists on a nonlinear @xmath1 crystal placed inside an optical cavity ; when pumped above some threshold level with a laser beam at frequency @xmath2 , a field oscillating at half that frequency ( _ signal _ frequency ) is generated . a linearized analysis of the dopo s quantum fluctuations reveals that the phase quadrature of the signal field can be perfectly squeezed when pump s power is tuned to the dopo s threshold value @xcite . of course , ideal perfect squeezing can not be real in this case , as it would entail an infinite number of photons in the generated mode , and it can be shown that nonlinear corrections make this squeezing level become finite @xcite . apart from fundamental reasons , improving
the quality of squeezed light is an important task because of its applications .
for example , in the fields of _ quantum information with continuous variables _ @xcite ( as mixing squeezed beams with beam splitters offers the possibility to generate multipartite entangled beams @xcite ) and _ high _ _ precision measurements _ ( such as beam displacement and pointing * * measurements @xcite or gravitational wave detection @xcite ) , applications of squeezed light have been theoretically and experimentally proved .
recently , our group has developed a strategy that would lead to the generation of light with high level of squeezing at any pump level above threshold in dopos ( _ non - critically squeezed light _ ) .
the idea is to allow for the existence of several transverse modes at the signal frequency in a dopo possessing some spatial symmetry in the transverse plane ( e.g. , rotational or translational ) ; under some circumstances , a pattern breaking the corresponding spatial symmetry can be generated , and quantum noise can randomly move it along the invariant direction ( i.e. , rotate or translate it in the transverse plane ) .
this can be seen as an indefiniteness in the transverse position of the generated pattern , which invoking now the uncertainty principle , could be accompanied by the perfect determination of its associated momentum ( angular or linear ) .
hence , we could expect noise reduction in the empty pattern coinciding with the momentum of the generated pattern .
moreover , as this is related to the spontaneous symmetry breaking , the perfect squeezing should occur at any pump level above threshold . in ref .
@xcite , the translational symmetry breaking was considered in broad area , planar dopos , where cavity solitons ( css ) , as well as extended patterns , have been predicted to exist @xcite .
it was shown that all the reasoning above is true : the position of the cs diffuses in the transverse plane , and the phase quadrature of the pattern coinciding with its linear momentum ( namely its @xmath3 phase shifted transverse gradient ) is perfectly squeezed ( within the linearized theory ) at any pump level above threshold .
the problem of this model is that it is not too close to current experimental setups and even css have not yet been observed in dopos .
this was one of the reasons why the theory of a simpler system was developed in ref .
@xcite : a rotationally symmetric dopo tuned to the first family of transverse modes at the signal frequency @xcite .
the first transverse - mode family supports laguerre - gauss ( lg ) modes @xmath4 with @xmath5 orbital angular momentum ( oam ) , where @xmath6 are the transverse coordinates .
when pumped with a gaussian mode with zero oam , two signal photons with opposite oam are generated in the @xmath1 crystal .
alternatively , the simultaneous generation of one @xmath7 photon and another @xmath8 one corresponds to the generation of two hermite - gauss ( hg ) tem@xmath0 photons , the orientation of this tem@xmath0 mode ( @xmath9 in fig .
1 ) given by half the phase difference between the subjacent @xmath10 modes . again in this case , we proved our reasoning given above : quantum noise is able to rotate randomly this bright mode , and the phase quadrature of its angular momentum ( its @xmath11 phase shifted angular derivative ) , which corresponds in this case to another hg mode orthogonal to the generated one ( we will refer to it as the dark mode ) , is perfectly squeezed within the linearized theory and at any pump level above threshold @xcite .
in addition we showed that this result is quite robust against deviations from the perfect rotational invariance of the dopo .
dopo.eps + figure 1.- scheme of the 2-transverse - mode dopo .
moreover , a surprising result was found : although the phase quadrature of the dark mode is perfectly squeezed , its corresponding amplitude quadrature has not increased uncertainty , i.e. , the quadratures of this rotating mode do not form a canonical pair .
we intuitively explained this by noting that the excess of noise is transferred not to another quadrature , but to the orientation @xmath9 of the pattern , which actually is the canonical pair of the squeezed quadrature ; in the present article we prove this statement by using the arguments developed in ref .
@xcite .
indeed , the study of this rotationally symmetric dopo was carried out under some assumptions whose repercussion will be analyzed in the current article .
first , we supposed that cavity losses at the signal frequency were small as compared with that at the pump frequency , what allowed us to adiabatically eliminate the pump variables .
next , the usual linearization of the nonlinear langevin equations around their classical stationary solution was done . finally , in order to prove the non - critical squeezing we proposed an homodyne detection scheme in which the local oscillator was a tem@xmath0 mode matched to the dark mode at any time
, i.e. , it was following the random rotation of this mode , an ideal situation which is not possible in real experiments .
now , let us summarize what we are going to show in this article , which is divided into four main parts .
in the first one ( section ii ) we describe the dopo tuned to the first transverse family ( the _ 2-transverse - mode dopo _ ) and find its associated quantum langevin equations . in section iii , we study the classical emission of the system , and show how the rotational symmetry is broken by the generated transverse pattern .
section iv is the main section of the article : the quantum properties of the dopo are analyzed by linearizing the langevin equations . here
we will prove the random rotation of the classical pattern , and find the squeezing properties of the bright and dark modes .
in addition , we will show that the quadratures of the dark mode do not form a canonical set , but they are canonically related to the pattern orientation ( where the excess of noise goes ) .
we close this section by analyzing the repercussion of keeping fixed the orientation of the local oscillator , showing that even in this case large levels of squeezing can be obtained . in the last part ( section v )
we show that the approximations considered in the previous sections ( the adiabatic elimination of the pump and the linearization procedure ) are not responsible for the remarkable squeezing properties of the dark mode , as they follow directly from the spontaneous rotational symmetry breaking . at the end ( section vi ) we will give some conclusions . in order to make clearer the physics behind the phenomenon , we have left most of the technical details to appendices .
one of the simplest configurations of the dopo s cavity allowing for the generation of a rotationally asymmetric pattern is the following @xcite : at some phase - matched frequency @xmath2 it is tuned to a fundamental gaussian transverse mode @xmath12 , while at half that frequency , @xmath13 , the first transverse family is resonant . in the first transverse family
two laguerre - gauss modes exist , @xmath4 . the exact expression of the modes in the waist plane of the resonator ( where the @xmath1 crystal is placed ) is given by @xcite@xmath14 where @xmath15 is the coordinate vector in the transverse plane , and @xmath16 is the beam radius at the waist plane , which depends on the cavity geometry and the resonance frequency of the mode ( see appendix a ) .
these transverse modes have @xmath17 and @xmath5 oam respectively , as they are eigenmodes of the oam operator @xmath18 with those eigenvalues .
the field inside the cavity can be written then as@xmath19 with@xmath20 being @xmath21 the refractive index of the crystal and @xmath22 the effective cavity length ( h.c
. stands for hermitian conjugate ) .
the slowly varying envelopes are given by [ sve]@xmath23 where the boson operators satisfy the usual equal time commutation relations @xmath24 = \delta_{mn};\text { } m , n=0,\pm1.\ ] ] instead of using the lg basis , one can work in the most usual tem@xmath25 ( hg ) basis .
denoting by @xmath26 and @xmath27 a tem@xmath0 mode rotated an angle @xmath28 with respect to the @xmath29 axis and its orthogonal , these are given by [ hg]@xmath30 \label{hg10}\\ & = \sqrt{2}\left\vert l_{\pm1}\left ( \mathbf{r}\right ) \right\vert \cos\left ( \phi-\psi\right ) \nonumber\\ h_{01}^{\psi}\left ( \mathbf{r}\right ) & = \frac{1}{\sqrt{2}i}\left [ e^{-i\psi}l_{+1}\left ( \mathbf{r}\right ) -e^{i\psi}l_{-1}\left ( \mathbf{r}\right ) \right ] \label{hg01}\\ & = \sqrt{2}\left\vert l_{\pm1}\left ( \mathbf{r}\right ) \right\vert \sin\left ( \phi-\psi\right ) .\nonumber\end{aligned}\ ] ] one can thus define boson operators associated to these modes , whose relation with the lg ones is [ lgtohg]@xmath31 these relations will be important in order to understand the properties of the system . finally , note that for any mode @xmath32 , quadrature @xmath33 can be defined as @xmath34 and quadratures with @xmath35 and @xmath36 are usually called amplitude and phase quadratures , and will be denoted by @xmath37 and @xmath38 , respectively .
once the structure of the field inside the cavity has been described , we can pass to describe its evolution , and to this aim a hamiltonian must be built .
it has to take into account two processes : the pumping of the coherent , gaussian , resonant laser source at frequency @xmath2 , and the parametric down conversion of the pump photons into signal photons occurring in the @xmath1 crystal . in the interaction picture , this hamiltonian
is given by@xmath39 where @xmath40 and @xmath41 are proportional to the injected pump amplitude and to the second order nonlinear susceptibility of the nonlinear crystal respectively .
explicit expressions of these parameters in terms of physical quantities are given in appendix a. we take @xmath40 as real , i.e. , we take the phase of the pump laser as the reference phase for all the other fields .
the first part of the hamiltonian is the usual pump hamiltonian @xcite , while the second part is easily justified by energy and oam conservation : from one @xmath2 photon with zero oam , two @xmath42 photons are created with opposite oam , or viceversa .
note that this hamiltonian is equivalent to that for the nondegenerate optical parametric oscillator considered for example in refs .
@xcite .
there exists one more process not taken into account in this hamiltonian : the losses through the cavity mirrors .
this is an irreversible process which can not be described with a hamiltonian formalism that takes into account only the cavity dynamics .
however , assuming that the outer of the cavity consists of a continuum of modes in a vacuum ( or coherent ) state which weakly interact with the internal modes through the partially reflecting mirror , it can be incorporated into the master equation satisfied by the density operator of the system as shown for example in ref .
moreover , the master equation can be converted into a set of stochastic ( langevin ) differential equations by using a positive _ p _ representation for the density operator @xcite . in our case , as is shown in @xcite , the equivalent set of langevin equations is@xmath43
^{\ast},\nonumber\end{aligned}\ ] ] where @xmath44 and @xmath45 are independent complex amplitudes , @xmath46 and @xmath47 are the decay rates of the cavity at the pump and signal frequencies respectively ( see appendix a ) , and the independent complex noises @xmath48 and @xmath49 have zero mean and nonzero correlations@xmath50 ^{\ast}\right\rangle = \delta\left ( t_{1}-t_{2}\right ) .
\label{noisecorr}\ ] ] the equivalence between the master equation of the system and these equations must be understood in the following way:@xmath51 i.e. , quantum expected values of normally ordered functions equal stochastic averages of the same functions changing boson operators @xmath52 by independent complex stochastic variables @xmath53 .
note that these stochastic equations are equal in either the ito or stratonovich forms ( see appendix b ) , and hence we interpret them as stratonovich equations which allow us to apply ordinary calculus . in order to better visualize the free parameters of the model ,
let us define new rescaled time and amplitudes through@xmath54 and similar expressions for the conjugate fields @xmath55 and @xmath56 .
note that as time is measured in units of @xmath57 , the noises must be rescaled also as@xmath58 and similarly for @xmath59 , if we want to preserve their statistical properties ( [ noisecorr ] ) in terms of the adimensional time @xmath60 .
finally , these changes make langevin equations ( [ langevin ] ) read ( of course , derivatives are respect with the new dimensionless time)@xmath61 ^{\ast},\nonumber\end{aligned}\ ] ] which have only the following 3 dimensionless parameters:@xmath62
before analyzing the quantum properties of the system , let us examine the emission of this dopo as predicted by classical optics .
it is possible to retrieve the classical equations of the 2-transverse - mode dopo from the quantum langevin ones ( [ reescaledlangevin ] ) by setting all the noises to zero and making @xmath63 ( @xmath32 refers to any mode of the problem ) .
we are interested in finding the stationary emission , and then set to zero the time derivatives of the amplitudes , arriving to@xmath64 before solving these equations , let us stress that these are invariant under the following transformation@xmath65 and hence the phase difference between opposite oam modes is classically undetermined . by decomposing into modulus and
phase the amplitudes @xmath66 , it is straightforward to find the solutions of the system ( [ classicaleqs ] ) .
if in addition a linear stability analysis is made , it is easy to find the following result : * for @xmath67 the only stable solution is@xmath68 and hence the signal modes are off if the pump @xmath40 does nt exceed a threshold value @xmath69 . * on the other hand , if @xmath70 , the signal modes are switched on , and the only stable solution in this case is@xmath71 where , as stated above , @xmath9 is arbitrary .
if we substitute the last result for the signal modes in the expression of the corresponding slowly varying envelope ( [ sves ] ) , we find the generated pattern to be@xmath72 = \sqrt{2}\rho h_{10}^{\theta}\left ( \mathbf{r}\right ) \text{,}\ ] ] which is a tem@xmath0 mode forming an angle @xmath9 with respect to the @xmath29 axis ( [ hg ] ) as shown in fig . 1 . due to the arbitrariness of @xmath9
, the pattern can arise with any orientation ; this reflects the rotational invariance of the system , wich in turn is broken after the tem@xmath0 mode generation . in the following ,
we will call @xmath73 the _ bright mode _ ( as it is classically excited ) and its orthogonal mode @xmath74 the _ dark mode _ ( as it is classically empty of photons ) , and will define the collective indices @xmath75 and @xmath76 to simplify the notation .
we are going to discuss the quantum properties of the down converted field by inspection of the quantum langevin equations ( [ reescaledlangevin ] ) in the limit @xmath77 , i.e. , @xmath78 , where the pump variables can be adiabatically eliminated .
we will show later ( section v ) that all the important properties found in this limit are valid in general .
one common way of treating the adiabatic elimination of the pump consists on setting @xmath79 in the first two equations of ( [ reescaledlangevin ] ) . however , this method has a problem : the initial langevin equations are equal in either ito or stratonovich forms , but the new langevin equations obtained for the signal modes does nt have this property ( see appendix b ) .
hence it is fair to ask within which interpretation ( ito or stratonovich ) is that procedure correct , if it is correct at all . as proved in @xcite by performing the adiabatic elimination in the fokker - planck equation ( where there are no problems of interpretation )
, the usual method for eliminating the pump is correct within ito s interpretation , and hence the stratonovich form of the stochastic equations satisfied by the signal modes only has an unexpected extra term proportional to @xmath80 ( see appendix b ) and reads:@xmath81 ^{\ast},\nonumber\end{aligned}\ ] ] with@xmath82 in order to find analytic predictions from these equations , we assume well above threshold operating conditions , where classical emission dominates , thus allowing for a linearization procedure
. the correct way to perform this was described in @xcite , and relies on the smallness of the @xmath83 parameter ( see appendix a ) . in particular , making a perturbative expansion of the amplitudes in terms of this parameter , it is possible to show that the classical equations are recovered as the @xmath84 order of the expansion in eqs .
( [ adeleqs ] ) , while the linear quantum correction appears in the @xmath85 order ( note that the noise term is already of order @xmath85 ) .
hence , the usual linearization procedure begins by writing the amplitudes as @xmath86 and @xmath87 , and treat the fluctuations as order @xmath83 perturbations .
however , as stated in the introduction , in the system we are dealing with we expect that quantum noise rotates the generated tem@xmath0 mode , and hence fluctuations of the fields in an arbitrary direction of phase space could not be small ( i.e. , order @xmath83 )
. nevertheless , eqs .
( [ adeleqs ] ) can be linearized if the amplitudes are expanded as@xmath88 e^{\mp i\theta\left ( \tau\right ) } \label{linexp}\\ \beta_{\pm1}^{+}\left ( \tau\right ) & = \left [ \rho+b_{\pm1}^{+}\left ( \tau\right ) \right ] e^{\pm i\theta\left ( \tau\right ) } , \nonumber\end{aligned}\ ] ] because as we will prove @xmath89 carries the larger part of the fluctuations , while the @xmath90 s and @xmath91 remain as order @xmath83 quantities .
in addition , expanding the fields in this way allows us to track the evolution of the classical pattern s orientation , as we take @xmath9 as an explicit quantum variable .
then , writing eqs .
( [ adeleqs ] ) up to order @xmath83 , we arrive to the following linear system ( arriving to this expression is not as straightforward as it might seem , see appendix c)@xmath92 with@xmath93 ^{\ast}\end{pmatrix } , \ ] ] and where @xmath94 is a real , symmetric matrix given by@xmath95 with the following eigensystem@xmath96{ll}\lambda_{0}=0 , & \mathbf{w}_{0}=\frac{1}{2}\operatorname{col}\left ( 1,-1,-1,1\right ) \\
\lambda_{1}=-2 , & \mathbf{w}_{1}=\frac{1}{2}\operatorname{col}\left ( 1,1,-1,-1\right ) \\ \lambda_{2}=-2\left ( \sigma-1\right ) , & \mathbf{w}_{2}=\frac{1}{2}\operatorname{col}\left ( 1,1,1,1\right ) \\
\lambda_{3}=-2\sigma , & \mathbf{w}_{3}=\frac{1}{2}\operatorname{col}\left ( 1,-1,1,-1\right ) .
\end{array } \label{eigensystem}\ ] ] defining the projections @xmath97 , and projecting the linear system ( [ linlan ] ) onto these eigenmodes , we find the following set of decoupled linear equations ( @xmath98 is set to zero , as otherwise it would just entail a redefinition of @xmath9 ) [ projlinlan]@xmath99 where the following real noises have been defined @xmath100 which have zero mean and nonzero correlations@xmath101 note finally that the solutions for the projections @xmath102 will be of order @xmath83 ( see appendix d ) , and hence so will be the @xmath90 s ( note that this is not the case for @xmath9 , whose initial value is completely arbitrary , although its variation @xmath91 is indeed of order @xmath83 ) .
this is consistent with the initial assumptions about the orders in @xmath83 of the involved quantities .
the first quantum effect that we are going to show concerns the orientation of the classically generated mode . equation ( [ thetaevo ] ) defines a _ wiener process _ for @xmath9 , thus showing that the orientation of this bright mode diffuses with time ruled by quantum noise .
how fast this diffusion is can be measured by evaluating the variance of @xmath9 . using the statistical properties of noise ( [ realnoisestat ] ) , it is straightforward to obtain the following result@xmath103 where @xmath104 with @xmath105 in ( [ theta var ] ) the notation @xmath106 has been used .
eq . ( [ theta var ] ) shows that the delocalization of the pattern s orientation increases as time passes by , though for typical system parameters ( see appendix a ) one finds @xmath107 , and hence the rotation of the pattern will be fast only when working terribly close to threshold @xcite . * definition and criterium for squeezing . *
_ _ in the introduction we defined squeezed light as that having some quadrature fluctuating below the vacuum level .
hence , in order to find out whether a light beam is in a squeezed state or not , one has to measure its _ quadrature _ _ fluctuations _ and then compare them with the value set by the vacuum state .
the question then is : what quantity accounts for quadrature fluctuations of a light beam ?
of course , in the case of a single mode of light this quantity can be directly the uncertainty of the quadrature .
however , outside the cavity there exists a continuum of modes , and the quantity accounting for these fluctuations has to be adapted to what can be most easily observed in an experiment . as was first shown in @xcite , the quantity accounting for fluctuations of quadrature @xmath33 outside the cavity ( @xmath32 refers to any signal spatial mode of our 2-transverse - mode dopo ) is@xmath108 with @xmath109 where the factor @xmath110 appears after including our rescaled variables ( [ rescaled ] ) .
we will call @xmath111 the _ noise spectrum _ and @xmath112 the _ squeezing spectrum_. frequency @xmath113 is usually called _ noise frequency _ , and it must not be confused with the optical frequency . in fact , noise frequency @xmath113 has contributions of every pair of modes lying in opposite sidebands around the optical frequency @xmath114 , where @xmath13 is the carrier frequency of the detected beam ( do nt miss that in our case this frequency is measured in units of @xmath47 ) .
@xmath111 can be measured via a balanced homodyne detection experiment : a coherent , intense field ( local oscillator field ) prepared in mode @xmath32 , and with a phase @xmath115 is mixed with the beam exiting the dopo in a 50/50 beam splitter ; it is easy to show that the operator associated to the intensity difference between the two output ports of the beam splitter is proportional to the quadrature @xmath116 of the beam exiting the dopo .
hence , the normalized correlation spectrum of the intensity difference ( which can be measured with a simple spectrum analyzer ) coincides with expression ( [ vout ] ) .
factor 2 in ( [ susual ] ) comes from the relation between the intracavity and the output modes when the input modes ( the pump laser and the outer modes ) are coherent ; this factor is @xmath117 when time ( frequency ) is not normalized to @xmath118(@xmath47 ) . in the above expression for the squeezing spectrum
, it is assumed that the emission of the dopo reaches a stationary state ( i.e. , the correlation function of @xmath119 at two different times @xmath60 and @xmath120 depends only on the time difference @xmath121 ) and that the observation time , say @xmath122 , is large as compared with the coherence time of @xmath119 . in the 2-transverse - mode dopo
there exists an undamped quantity excited by quantum noise , @xmath9 , and hence , any quantity with explicit dependence on it will not arrive to a stationary state .
thus , under some circumstances we will need the expression of the squeezing spectrum measured when the field is not stationary , which was derived in @xcite . in that reference
it is shown that when the measurement extends during a time interval @xmath122 ( normalized to @xmath118 in our expressions ) the squeezing spectrum reads@xmath123 , \label{sgen}\ ] ] which obviously reduces to ( [ susual ] ) if the emission is stationary .
for the vacuum state @xmath124 , hence if @xmath125 for some frequency @xmath126 , we can say that the detected light is in a squeezed state for mode @xmath32 at noise frequency @xmath126 .
finally , let us stress that we will evaluate quantum correlations as those involved in ( [ sgen ] ) via stochastic correlations , as allowed by property ( [ n - st ] ) .
hence , in the following we will change operators by their equivalent stochastic variables within the positive _ p _ representation in all the definitions above , and their associated normally ordered correlations by the corresponding stochastic correlations . *
independent quadratures and non critical squeezing . * a first step towards analyzing the squeezing properties of the field within the framework presented above , is identifying a set of independent quadratures .
this is easy in our case , as the eigenmodes @xmath127 give us a set of quadratures with well defined squeezing properties .
in particular , from ( [ linexp ] ) , ( [ lgtohg ] ) , and ( [ quadrature ] ) , it is easy to find the following relations@xmath128 where @xmath129 and @xmath130 are the amplitude and phase quadratures of the bright and dark modes , @xmath131 and @xmath74 , respectively . the evolution of these quadratures can be found from the equations satisfied by the projections ( [ projlinlan ] ) , which are solved in appendix d ( remember that @xmath132 ) .
in particular , there we show that after some time these projections reach a stationary state , and hence we can evaluate the noise spectrum of the quadratures by using the stationary expression for the squeezing spectrum ( [ susual ] ) . using the results found in appendix d for the correlation spectrum of the projections ( [ cspectrum ] ) ,
it is straightforward to find the following results [ noisespectra]@xmath133 we see that the quadratures of the bright mode have the same behavior as those of the single mode dopo : the phase quadrature @xmath134 is perfectly squeezed @xmath135 at zero noise frequency @xmath136 only at the bifurcation @xmath137 . on the other hand , the dark mode has perfect squeezing in its phase quadrature @xmath138 at zero noise frequency .
what is interesting is that this result is independent of the distance from threshold , and thus , it is a non - critical phenomenon .
this result was first shown in @xcite , and we can understand it by following the reasoning given in the introduction : as the orientation @xmath9 of the classically excited pattern is undetermined in the long - time limit , its oam must be fully determined at low noise frequencies . on the other hand ,
the oam of the bright mode @xmath73 is nothing but its @xmath3 phase shifted orthogonal hg mode , i.e. , @xmath139 , which used as a local oscillator in an homodyne detection experiment would lead to the observation of the @xmath138 quadrature fluctuations . * on canonical pairs and noise transfer*. although all the results we have discussed up to now were expected to occur after the arguments given in the introduction , a strange unexpected result has appeared in ( [ noisespectra ] ) : the quadratures of the dark mode seem to violate the uncertainty principle as @xmath140 . moreover , the noise spectrum of an arbitrary quadrature of the dark mode , which can be written in terms of its amplitude and phase quadratures as @xmath141 , is @xmath142 what shows that all the quadratures of the dark mode ( except its amplitude quadrature ) are squeezed .
thus two orthogonal quadratures can not form a canonical pair as they satisfy the relation @xmath143 in clear violation of the uncertainty principle .
the natural question now is : where does the excess of noise go if it is not transferred from one quadrature to its orthogonal one ?
the answer is that it goes to the pattern orientation , which is actually fully undetermined in the long term as we showed above ( [ theta var ] ) .
this section is devoted to prove this statement .
in particular we will prove that two orthogonal quadratures of the dark mode do not form a canonical pair , while the orientation @xmath9 is the canonical pair of all the squeezed quadratures .
one way to prove this would be to evaluate the commutator between the quantum operators involved .
however , @xmath9 is half the phase difference between the opposite oam modes @xmath144 , whose associated operator has a very difficult expression @xcite , making the calculation of the needed commutators quite hard .
nevertheless , in @xcite we developed a simple proof based on classical field methods , and this is the one we present here .
a usual approach one uses to move from classical to quantum optics is to change the classical normal variables for each mode of the field , @xmath66 and @xmath145 , by boson operators @xmath146 and @xmath147 , satisfying commutation relations@xmath148 = i\left\ { \beta_{m},\beta_{m}^{\ast}\right\ } , \ ] ] where @xmath149 denotes the _ poisson bracket _ between two functions @xmath150 and @xmath151 defined as@xmath152 as an example , the poisson bracket of two monomode orthogonal quadratures @xmath153 and @xmath154 is found to be@xmath155 in general two phase space functions @xmath156 and @xmath157 are said to form a canonical pair if their poisson bracket is of the form@xmath158 with @xmath159 a real number . in our case , the functions we are interested in are the classical counterparts of the dark mode quadratures , which using ( [ lgtohgs ] ) with @xmath160 are written as @xmath161 + \mathrm{c.c.},\ ] ] with the orientation given by@xmath162 .\ ] ] now using the definition of the poisson brackets with @xmath163 and after some algebra , it is possible to show that@xmath164 and@xmath165 with @xmath166 . on the other hand , in the two - transverse - mode dopo , the number of photons with opposite oam is sensibly equal , i.e. , @xmath167 ; hence , the dominant term of the previous brackets will be that with @xmath168 , and thus@xmath169
and@xmath170 comparing these with ( [ canbrarel ] ) we find that this is a confirmation of what we expected : two orthogonal dark quadratures @xmath171 and @xmath172 do not form a canonical pair ( moreover , they commute ) , while @xmath171 and @xmath9 do .
this can be seen as an indirect proof of the same conclusion for the corresponding operators . * homodyne detection with a fixed local oscillator*.
so far we have considered the situation in which one is able to detect independently the bright and dark modes . however , as shown by eq .
( [ thetaevo ] ) , these modes are rotating randomly , what means that a local oscillator field following that random rotation should be used in order to detect them separately .
this might be a really complicated , if not impossible , task , so we analyze now the more realistic situation in which the local oscillator is matched to the orthogonal orientation of the emerging pattern only at some initial time , remaining then with the same orientation during the observation time .
we will show that even in this case , and as the rotation of the modes is quite slow ( [ theta var ] ) , large levels of noise reduction can be obtained . without loss of generality , we suppose that the bright mode emerges from the resonator at some initial time @xmath173 oriented within the @xmath29 axis , i.e. , @xmath174 .
hence , by using a tem@xmath175 local oscillator with a phase @xmath115 , the quadrature @xmath176 of a fixed @xmath177 mode ( the initially dark mode ) will be measured . in terms of the gauss - laguerre modes ,
the amplitude @xmath178 of the this mode is given by ( [ lgtohgs ] ) with @xmath179 .
then , by using the expansion ( [ linexp ] ) of the amplitudes @xmath66 as functions of the fluctuations @xmath180 and the orientation @xmath9 , quadrature @xmath176 of this mode can be rewritten as@xmath181 hence , the 2-time correlation function of @xmath176 yields ( note that ( [ projlinlan ] ) shows that @xmath182 , @xmath183 , @xmath184 , @xmath185 and @xmath186 are uncorrelated)@xmath187 \label{stillcorrelation}\\ & \text { \ \ } + 2\sin^{2}\varphi\left [ c_{1}\left ( \tau_{1},\tau_{2}\right ) c\left ( \tau_{1},\tau_{2}\right ) -c_{3}\left ( \tau_{1},\tau_{2}\right ) s\left ( \tau_{1},\tau_{2}\right ) \right ] , \nonumber\end{aligned}\ ] ] with@xmath188 as shown in appendices d and e , the latter correlation functions can be evaluated by using the linear evolution equations of the projections @xmath189 and @xmath9 -see eqs .
( [ ccorr ] ) , ( [ sincorr ] ) and ( [ coscorr])- ; then from ( [ stillcorrelation ] ) the squeezing spectrum of @xmath190 can be found by using the general expression ( [ sgen ] ) , as in this case the stationary expression ( [ susual ] ) can not be used because @xmath191 and @xmath192 do nt reach a stationary state .
although the general expression for @xmath193 is too lengthy to be written here , a more compact approximated expression can be found in the limit of small @xmath194 , leading to the following expression for the noise spectrum@xmath195
with@xmath196 and@xmath197 } { \left ( \sigma-1\right ) \left ( 4+\omega^{2}\right ) \left ( 4\sigma^{2}+\omega^{2}\right ) } , \ ] ] where @xmath198 . in the following
we will fix @xmath199 to @xmath200 ( pump power twice above threshold ) , as the results are almost independent of its value as far as it is far enough from threshold .
hence , the free parameters will be the detection parameters @xmath122 , @xmath113 and @xmath115 , and the diffusion @xmath194 which depends on the system parameters .
fig2.eps + figure 2.- zero - frequency noise spectrum of the phase quadrature corresponding to a fixed tem@xmath175 mode as a function of the detection time @xmath122 .
three different values of @xmath194 are considered ( @xmath201 , @xmath202 and @xmath203 from top to bottom ) .
the inset shows also this spectrum at zero noise frequency , but evaluated at the optimum detection time @xmath204 and as a function of @xmath194 ( note that the @xmath194 axis is in logarithmic scale ) . as mentioned in the text @xmath205 . in this section ,
the results for @xmath111 are presented in units , defined through the relation @xmath206
= 10\log v^{\mathrm{out}}$ ] ( hence , e.g. , @xmath207 and @xmath208 correspond to 90% of noise reduction ( @xmath209 ) and complete noise reduction ( @xmath210 ) respectively . ) from expression ( [ fixednoisespectrum ] ) we see that the maximum level of squeezing is obtained at @xmath211 ( see also fig . 3a ) and when the phase of the local oscillator is tuned exactly to @xmath3 . in fig .
2 we show the noise spectrum ( [ fixednoisespectrum ] ) for these parameters as a function of the detection time @xmath122 for 3 different values of @xmath194 .
we see that in all cases there exist an optimum detection time for which squeezing is maximum . minimizing eq .
( [ fixednoisespectrum ] ) with @xmath211 and @xmath36 with respect to @xmath122 , it is straightforward to find that this optimum detection time is given by@xmath212 with an associated noise spectrum @xmath213 ( shown in the inset of fig .
2 as function of @xmath194 ) .
these results show that large levels of noise reduction are obtained for the phase quadrature of the fixed tem@xmath175 mode , even for values of the diffusion parameter @xmath194 as large as @xmath214 ( remember that @xmath203 is a more realistic value ) .
however , in real experiments it is not possible to ensure that @xmath215 with an uncertainty below approximately @xmath216 @xcite , and hence we proceed now to investigate the level of noise reduction predicted by ( [ fixednoisespectrum ] ) when the local oscillator phase is different from @xmath215 .
fig3.eps + figure 3.- ( a ) noise spectrum of the fixed tem@xmath175 mode as a function of the noise frequency @xmath113 , and for four different values of the local oscillator phase ( @xmath217 , @xmath218 , @xmath219 , and @xmath220 from top to bottom curves ) .
it can be appreciated how the infinite fluctuations of @xmath221 at zero noise frequency enter in the spectrum for any @xmath222 .
the rest of parameters are @xmath205 , @xmath223 , and @xmath224 , but the same behavior appears for any election of the parameters .
( b ) noise spectrum of the fixed tem@xmath175 mode evaluated for the optimum parameters @xmath225 and @xmath204 as a function of @xmath115 ( @xmath226 for the blue - solid curve and @xmath214 for the red - dashed one , having both @xmath205 ) .
in addition , it is plotted the analogous curve for the single - mode dopo ( gold , dashed - dotted curve ) .
the inset shows the dependence of the optimum frequency @xmath225 with the phase of the local oscillator @xmath115 .
of course , when @xmath227 the noise frequency with maximum squeezing is no longer @xmath211 , as in this case the infinite fluctuations of @xmath228 at zero noise frequency , due to the rotation noise , enter the noise spectrum ( see fig .
3a ) . by minimizing @xmath111 with respect to @xmath113 and @xmath122 for different values of @xmath115 and @xmath194 ,
it is possible to show that the optimum value of the detection time is almost independent of @xmath115 for small deviations of this from @xmath220 , and hence it is still given to a good approximation by ( [ topt ] ) , though in this case this minimum is less pronounced than in the @xmath215 case shown in fig .
2 ( i.e. , the curve is almost horizontal around @xmath204 ) . on the other hand ,
the optimum noise frequency @xmath225 is independent of @xmath194 and depends on @xmath115 as shown in the inset of fig . 3b . as for the squeezing level , in fig .
3b we show the noise spectrum evaluated at @xmath204 and @xmath225 as a function of @xmath115 for 2 different values of the diffusion @xmath194 . together with these curves ,
we have plotted the noise spectrum of the single - mode dopo @xcite evaluated for its optimum parameters ( in this case it is optimized respect to @xmath199 and @xmath113 ) as a function of @xmath115 .
we see that the noise reduction is independent of @xmath194 as @xmath115 is taken apart from @xmath220 . on the other hand ,
the squeezing level is similar to that of the single - mode dopo , as the maximum difference between them are 1.5 ( a factor 1.4 in the noise spectrum ) in favor of the single - mode dopo , with the advantage that in the 2-transverse - mode dopo this level is independent of the distance from threshold .
therefore we see that the phenomenon of non - critical squeezing through spontaneous rotational symmetry breaking could be observed in the 2-transverse - mode dopo without the need of following the random rotation of the generated pattern , which makes its experimental realization feasible with current available technology .
the first assumption made in the search for the quantum properties of the system was that @xmath77 , a limit that allowed the adiabatic elimination of the pump field .
now we are going to show analytically that the phenomenon of squeezing induced by spontaneous rotational symmetry breaking is still present even without this assumption , but still working in the domain where linearization is correct . the way to show this is quite simple ; starting from the complete equations ( [ reescaledlangevin ] ) , we expand the amplitudes @xmath66 around the classical stationary solution ( [ abovethreshold ] ) as we made in ( [ linexp ] ) , but adding now a similar expression for the pump amplitudes : @xmath229 and @xmath230 ( note that for the pump modes the @xmath90 s are directly small as the phase of this mode is locked to that of the injection @xmath40 ) . then , linearizing these equations for the fluctuations and noises , we arrive to a linear system formally equal to ( [ linlan ] ) , but with@xmath231
^{\ast}\end{pmatrix } , \ ] ] and a linear matrix@xmath232 although in this case @xmath94 is not hermitian , it can be checked that it possess a biorthonormal basis @xcite , and in particular , the following two vectors are present in its eigensystem : @xmath233 and @xmath234 , with corresponding eigenvalues @xmath235 and @xmath236 .
these eigenvectors have null projection onto the pump subspace , and coincide with @xmath237 and @xmath238 in what concerns to the signal subspace ( [ eigensystem ] ) .
hence , all the properties derived from these vectors are still present without any change .
in particular , as they account for the diffusion of @xmath9 and the squeezing properties of the dark mode , we can conclude that these properties are still present when working out of the limit @xmath239 .
in this section we will show that the diffusion of the orientation and the associated non - critical squeezing of the dark mode , which have been found by linearizing the langevin equations , are also present when we consider the full nonlinear problem .
to do so , we will solve numerically the complete stochastic equations ( [ reescaledlangevin ] ) using the semi - implicit algorithm developed by drummond and mortimer in @xcite .
the details of the numerical simulation are explained in appendix f. here we just want to point out that the important parameters of the simulation are the step size @xmath240 used to arrive from @xmath173 to the final integration time @xmath241 , and the number of stochastic trajectories , say @xmath242 , which are used to evaluate stochastic averages .
the initial conditions @xmath243 are not important as the results in the stationary limit are independent of them .
the system parameters which have been chosen for the simulation are @xmath205 , @xmath244 ( to show also numerically that the adiabatic elimination has nothing to do with the phenomenon ) , and @xmath245 . we havent chosen a smaller value for @xmath83 ( like @xmath214 as followed from the physical parameters considered in appendix a ) because such a small number can make the simulation fail ; nevertheless all the results we are going to show should be independent of @xmath83 and @xmath199 , and we have also tested that the same results are obtained for other values of these .
fig4.eps + figure 4.- ( a ) evolution of the variance of the orientation @xmath9 given by the numerical simulation .
it has been divided by the slope @xmath246 predicted by the linearized theory ( [ theta var ] ) , so the straight line obtained is in perfect agreement with this linear result .
quantitatively , a linear regression analysis shows that the slope obtained by the data is @xmath247 , with standard error @xmath248 .
( b ) noise spectrum of the amplitude ( red - dashed curve ) and phase ( blue - solid curve ) quadratures of the dark mode as obtained by the numerical simulation .
the results are in perfect agreement with the ones predicted by the linearized theory ( [ noisespectra ] ) . a nonlinear regression analysis of the numerical data respect to the function @xmath249 $ ] , shows that @xmath250 are the best fit parameters having both standard errors @xmath251 , which is in good agreement with the values @xmath252 predicted by ( [ noisespectra ] ) .
it is also important to note that we have defined a general quadrature of the rotating dark mode as ( directly from ( [ lgtohgs ] ) with @xmath253)@xmath254 \\ & -\frac{i}{\sqrt{2}}\left [ e^{i\varphi}\left ( e^{-i\theta}\beta_{+1}^{+}-e^{i\theta}\beta_{-1}^{+}\right ) \right ] , \nonumber\end{aligned}\ ] ] with @xmath9 defined within the positive _ p _ representation through@xmath255 now let us show the results evaluated for the following simulation parameters : @xmath256 , @xmath257 and @xmath258 . this simulation has been compared with other ones having different values of these parameters to ensure convergence . in fig .
4a we show the variance of @xmath9 as a function of time .
the result has been normalized to @xmath246 , so that the linear result ( [ theta var ] ) predicts a straight line forming 45@xmath259 with respect to the time axis .
it can be appreciated that this is indeed what shows the simulation . in fig .
4b , we show the numerical results for the noise spectrum associated to the quadratures of the dark mode .
only times above @xmath260 have been considered in the correlation function to ensure being working in the stationary limit .
again , the results shown in fig .
4b are in perfect agreement with the linear predictions ( [ xsspectrum ] ) and ( [ ysspectrum ] ) .
these results show that the phenomenon of non - critically squeezed light via spontaneous rotational symmetry breaking is not a product of the linearization .
in conclusion , we have developed in detail the quantum theory of the 2-transverse - mode dopo introduced in @xcite .
we have studied some important features not considered in that reference : * when pumping with a gaussian mode a rotationally symmetric dopo properly tuned to the first family of transverse modes at the signal frequency , classical emission takes place in a tem@xmath0 mode with an arbitrary orientation in the transverse plane .
hence , once the threshold is crossed , the rotational symmetry is spontaneously broken , and we can talk about a bright mode ( the generated one which breaks the symmetry ) and a dark mode ( the mode orthogonal to the generated one ) . *
the symmetry breaking reflects in the appearance of a goldstone mode in the matrix governing the linear evolution of the fluctuations above threshold .
the null eigenvalue of this mode allows quantum noise to change the orientation of the bright mode randomly . though continuously increasing with time
, this rotation of the classically excited pattern is quite slow when working above threshold . * as for the squeezing properties , it has been proved that the bright mode has the same behavior as the single - mode dopo , i.e. , perfect squeezing appears only at threshold ( within the linearized theory ) and degrades fast as pump is moved apart from this level . on the other hand , accompanying the goldstone mode it appears another mode whose associated eigenvalue takes the minimum possible value @xmath261 .
these modes are responsible of the remarkable properties of the dark mode : its phase quadrature is perfectly squeezed at any pump level , while its amplitude quadrature carries only with vacuum fluctuations @xcite ( in apparent violation of the uncertainty principle ) .
a simple explanation of this phenomenon in terms of `` angle - angular momentum '' uncertainty relation appears once one notices that the dark mode coincides ( up to a @xmath3 phase ) with the oam of the generated pattern .
* we have proved that the apparent violation of the uncertainty principle is just that , apparent , as the conjugate pair of the squeezed quadrature is not another quadrature but the orientation of the bright mode , which in fact is completely undetermined in the long term . *
next we have pointed out that in order to measure the quantum properties of the dark mode , one has to use a tem@xmath0 local oscillator that is perfectly matched to the orientation of this mode at any time .
however , the mode is rotating randomly , which seems to make impossible the perfect matching .
for this reason we have studied the situation in which the local oscillator is matched to the dark mode s orientation only at the initial time , remaining fixed during the detection time .
we have shown that arbitrarily large levels of noise reduction can be obtained even in this case if the phase of the local oscillator is exactly @xmath3 .
we then considered phase deviations up to 2@xmath259 ( 1.5@xmath259 seems to be the current experimental limit @xcite ) , comparing the results with that predicted for the single - mode dopo ; similar levels are obtained for both , with the advantage that in the 2-transverse - mode dopo this level is independent of the distance from threshold , and hence , non - critical .
* in the last part of the article we have shown that the assumptions made in order to analytically solve the problem are not the responsible for the quantum properties of the dark mode . in particular , we have shown that when the pump is not adiabatically eliminated , the goldstone mode and its companion with the lowest possible eigenvalue remain unchanged in the matrix governing the linear evolution of the full problem , and hence all the properties derived from them are still present .
finally , we have used numerical simulations to show that the predictions of the linearized equations are in perfect agreement with those of the full nonlinear equations , and hence the perfect , non - critical squeezing of the dark mode is not a by - product of the linearization .
we believe that the analyses presented in this paper ( together with the fact that the properties of the dark mode are not too sensitive to imperfections in the rotational symmetry of the dopo @xcite , and that the phenomenon is present in other kinds of nonlinear resonators @xcite ) show that that the phenomenon of non - critical squeezing induced by spontaneous rotational symmetry breaking is a robust phenomenon .
we thank ferran v. garcia ferrer for his help in section v.a .
this work has been supported by the spanish ministerio de educacin y ciencia and the european union feder through project fis2008 - 06024-c03 - 01 .
c n - b is a grant holder of the fpu programme of the ministerio de educacin y ciencia ( spain ) .
a.r . acknowledges financial support from the universitat de valncia through its program convocatoria de estancias temporales para investigadores invitados " .
during the article some parameters have been used to model the dopo . here
we want to give explicit expressions for them in terms of physical quantities for the case of a dopo having a fabry - perot cavity of effective length @xmath22 and formed by two identical spherical mirrors with curvature radii @xmath262 for simplicity .
the nonlinear crystal is placed at the waist plane of this cavity and has refractive index @xmath21 , second order susceptibility @xmath263 and axial length @xmath264 ( assumed to be much smaller than the rayleigh length of the cavity ) . for this cavity configuration , the beam radius at the waist plane of the resonator is given by @xcite@xmath265 where @xmath266 is the wavelength of the considered mode inside the cavity .
on the other hand , the parameters which appear in the hamiltonian ( [ hamiltonian ] ) accounting for the external laser pump and the nonlinear interaction inside the crystal , namely @xmath40 and @xmath41 , respectively have the following expression:@xmath267 being @xmath268 the cavity decay rate at the considered frequency ( @xmath269 is the corresponding transmission factor through the input mirror ) , and @xmath270 the power of the injected laser . in order to obtain these expressions all
the transmitted pump power is assumed to be focalized inside the transverse dimensions of the nonlinear crystal . through the article ,
some expressions have been evaluated for concrete system parameters .
we have taken as typical parameters the following ones@xmath96{llll}\lambda_{\mathrm{p}}=400\text { } \operatorname{nm } , & r=1\text { } \operatorname{m } , & \mathcal{t}_{\mathrm{p}}=0.1 , & \chi^{\left ( 2\right ) } = 2\text { } \frac{\operatorname{pm}}{\operatorname{v}},\\ l=0.1\text { } \operatorname{m } , & l=1\text { } \operatorname{mm } , & \mathcal{t}_{\mathrm{s}}=0.01 , & n=2.5 , \end{array}\ ] ] leading to the following model parameters@xmath96{ll}w_{\mathrm{p}}=167\text { } \operatorname{\mu m } , & \chi=64\text { } \operatorname{s}^{-1},\\ \gamma_{\mathrm{p}}=0.15\text { } \operatorname{ns}^{-1 } , & \gamma_{\mathrm{s}}=0.015\text { } \operatorname{ns}^{-1}. \end{array}\ ] ]
during the article we make extensive use of the equivalence between langevin and fokker - planck equations . in this appendix
we want to briefly review this connection .
consider a set of real variables @xmath271 , satisfying stochastic langevin equations@xmath272 where @xmath273 and @xmath274 are a vector and a matrix which depend on the variables , and the components of @xmath275 are real noises satisfying the usual statistical properties ( [ realnoisestat ] ) .
the theory of stochastic processes @xcite states that the stochastic average of any function of the variables @xmath276 can be evaluated as@xmath277 where the probability distribution @xmath278 satisfies the fokker - planck equation@xmath279 p\left ( \tau;\mathbf{x}\right ) \text{,}\ ] ] having drift vector and diffusion matrix@xmath280 and@xmath281 respectively .
the parameter @xmath282 is @xmath17 or @xmath283 depending on whether we interpret eqs .
( [ genlan ] ) as ito or stratonovich stochastic equations . in the current article
any stochastic equation is interpreted la stratonovich , which allows us using the usual rules of calculus .
note that if the extra term involving derivatives of the noise matrix is zero , ito and stratonovich interpretations are equivalent , what happens for example on eqs .
( [ langevin ] ) but not on eqs .
( [ adeleqs ] ) .
following the linearization procedure we explained in section iii.a leads not to eq .
( [ linlan ] ) directly , but to the following one@xmath284 where@xmath285 and the rest of vectors and symbols were defined in the corresponding section .
note that the differences between this system of equations and the one used in the text ( [ linlan ] ) are the matrix @xmath286 and the @xmath287 term .
the latter is of order @xmath80 ( as the @xmath90 s and @xmath91 are of order @xmath83 ) , and hence it can be simply removed within the linearized theory .
understanding why @xmath286 can be removed from the linearized equations is a little more involved .
projecting these equations onto the eigensystem of @xmath94 ( [ eigensystem ] ) and defining the vector @xmath288 leads to the following system of equations ( remember that we set @xmath132 ) @xmath289 with@xmath290 being @xmath291 the 2-dimensional rotation matrix of angle @xmath9 acting on the @xmath292 subspace , and where the components of vector @xmath293 are real , independent noises satisfying the usual statistical properties ( [ realnoisestat ] ) .
now , we will prove that this system , and the same with @xmath294 are equivalent within the linearized theory , and hence ( [ fulllinlan ] ) and ( [ linlan ] ) are equivalent too . to show this
, we just write the fokker - planck equation corresponding to this stochastic system ( see appendix b ) , whose drift vector and diffusion matrix are found to be@xmath295 and@xmath296 respectively . note that in the last equation we have used that @xmath297 is an orthogonal matrix . the proof is completed by writing the stochastic system corresponding to this fokker - planck equation up to the linear order in @xmath83 , which reads@xmath298 corresponding to ( [ clinlan ] ) with @xmath294 as we wanted to prove .
hence , removing @xmath286 and neglecting the @xmath287 term from ( [ fulllinlan ] ) does nt change its equivalent fokker - planck equation within the linearized description , and thus equation ( [ linlan ] ) must lead to the same predictions as ( [ fulllinlan ] ) .
in this appendix we solve the linear evolution equations ( [ projlinlan ] ) of the projections @xmath189 , finding their 2-time - correlation function and the associated spectrum .
these equations are of the general type@xmath299 where @xmath300 is a real noise satisfying the usual white noise statistics ( [ realnoisestat ] ) , @xmath301 is a real , positive parameter , and @xmath302 is a parameter that might be complex . by making the variable change @xmath303 , and considering times larger that @xmath304 ( stationary limit ) , it is straightforward to find the following solution@xmath305 from the statistical properties of noise ( [ realnoisestat ] ) we see that this solution has zero mean , @xmath306 , and correlation@xmath307 considering separately the cases @xmath308 and @xmath309 , this integral is easily carried out , yielding ( again the limit @xmath310 is considered)@xmath311 where we see that this function depends only on the time difference @xmath312 , what justifies the name stationary limit for the @xmath313 approximation . finally , the spectrum of this correlation is found to be@xmath314 hence , particularized to the equations for the projections @xmath189 ( [ projlinlan ] ) , the correlations read@xmath315 while the spectra read@xmath316 and the stationary limit is reached when @xmath317 , @xmath318 and @xmath319 respectively .
in this appendix we show how @xmath320 and @xmath192 defined in ( [ stillcorr ] ) can be evaluated from the evolution equation of @xmath9 ( [ thetaevo ] ) .
as clearly explained in @xcite , this equation defines a wiener or random walk process equivalently described by the following fokker - planck equation:@xmath321 in the same reference , it is also proved that the 2-time joint probability associated to this simple diffusion equation is@xmath322 } , \ ] ] which gives us the probability of passing from orientation @xmath323 at time @xmath324 , to @xmath325 at time @xmath326 ( it is assumed that @xmath327 at the initial time @xmath173 ) .
of course , in this expression @xmath328 . from this joint probability the correlations are calculated as the integrals@xmath329 these integrals are easily evaluated , yielding@xmath330 , \label{sincorr}\ ] ] and@xmath331 .
\label{coscorr}\ ] ]
in this last appendix we want to briefly resume the details concerning the numerical simulation of the langevin equations ( [ reescaledlangevin ] ) .
the first important property of these equations is that , irrespective of the initial conditions , the amplitudes corresponding to opposite oam modes become complex - conjugate after a short transitory time , i.e. , @xmath332 ^{\ast}\right ) $ ] .
hence , if the initial conditions are chosen so that the oam pairs are complex - conjugate , we can be sure that they will remain complex - conjugate during the evolution .
in particular , we have chosen the above threshold stationary solution ( [ abovethreshold ] ) with @xmath327 as the initial condition . under these conditions , the 6 langevin equations ( [ reescaledlangevin ] ) reduce to the following 4 ( which we write in matrix form):@xmath333 with@xmath334 ^{\ast}\\ -\beta_{+1}^{+}+\beta_{0}^{+}\beta_{+1}^{\ast}\end{pmatrix } ,
\nonumber\\ \mathcal{b}\left ( \boldsymbol{\beta}\right ) & = g\operatorname{diag}\left ( 0,0,\sqrt{\beta_{0}},\sqrt{\beta_{0}^{+}}\right ) .\nonumber\end{aligned}\ ] ] in order to solve numerically these equations we use the semi - implicit algorithm developed in ref .
this algorithm is a finite - differences based method in which the total integration time @xmath241 ( the integration is supposed to begin always at @xmath335 ) is divided in @xmath336 segments , creating hence a lattice of times @xmath337 separated by time steps @xmath338 .
then , a recursive algorithm starts in which the amplitudes at time @xmath339 , say @xmath340 , are found from the amplitudes @xmath341@xmath342 at an earlier time @xmath343 from@xmath344 where @xmath345@xmath346 is an approximation to the amplitudes at the mid - point between @xmath343 and @xmath339 ( hence the name semi - implicit for the algorithm ) and the components of @xmath347 are independent discrete noises @xmath348 with null mean and satisfying the correlations@xmath349 ^{\ast}\right\rangle = \delta\tau\delta_{mn}\delta_{jk}.\ ] ] the mid - point approximation is found from the following iterative algorithm@xmath350 , \ ] ] where @xmath345@xmath351@xmath341@xmath342 , being @xmath352 the iteration index ( two iterations are carried in our simulations ) , while the discrete noises can be simulated at any step as @xcite@xmath353 , \ ] ] with@xmath354 being @xmath355 , @xmath356 , @xmath357 and @xmath358 independent random numbers uniformly distributed along the interval @xmath359 $ ] .
this algorithm allows us to simulate one stochastic trajectory .
then , by repeating it @xmath242 times , the stochastic average of any function can be approximated by the arithmetic mean of the values of that function evaluated at the different stochastic trajectories .
h. vahlbruch , m. mehmet , s. chelkowski , b. hage , a. franzen , n. lastzka , s. goszler , k. danzmann , and r. schnabel , phys .
* 100 * , 033602 ( 2008 ) ; see also m. mehmet , h. vahlbruch , n. lastzka , k. danzmann , and r. schnabel , arxiv : 0909.5386 ( 2009 ) .
n. treps , u. andersen , b. buchler , p.k .
lam , a. mitre , h.a .
bachor , and c. fabre , phys .
* 88 * , 203601 ( 2002 ) ; n. treps , n. grosse , w.p .
bowen , c. fabre , h .- a .
bachor , and p.k .
lam , science * 301 * , 940 ( 2003 ) .
h. vahlbruch , s. chelkowski , b. hage , a. franzen , k. danzmann , and r. schnabel , phys .
95 * , 211102 ( 2005 ) ; k. goda , o. miyakawa , e.e .
mikhailov , s. saraf , r. adhikari , k. mckenzie , r. ward , s. vass , a.j .
weinstein , n. mavalvala , nature physics * 4 * , 472 ( 2008 ) .
this result is linked to the noise reduction in the signal - idler intensity difference of a non - degenerate optical parametric oscillator @xcite . from a rigurous point of view , this is seen by writing a representation of the signal field s oam in terms of boson operators as @xmath360 -see eq . ( [ sves])- , showing that noise reduction in the intensity difference between opposite oam modes should entail squeezing in the oam of the generated pattern .
s. reynaud , c. fabre and e. giacobino , j. opt . soc .
b * 4 * , 1520 ( 1987 ) ; a.s .
lane , m.d .
reid and d.f .
walls , phys .
a * 38 * , 788 ( 1988 ) ; m.d .
reid and p.d .
drummond , phys .
a * 40 * , 4493 ( 1989 ) .
note that the quantitative difference between this difusion coeficient @xmath246 and that of @xcite , named @xmath361 in that reference , comes from the fact that @xmath246 is @xmath361 normalized to @xmath47 . c. navarrete - benlloch , g.j . de valcrcel , and e. roldn , phys .
a. * 79 * , 043820 ( 2009 ) ; f.v .
garcia - ferrer , c. navarrete - benlloch , g.j .
de valcrcel , and e. roldn , accepted in the ieee journal of quantum electronics ( 2009 ) . | in this article we explore the quantum properties of a degenerate optical parametric oscillator when it is tuned to the first family of transverse modes at the down converted frequency .
recently we found [ phys .
rev .
lett . * 100 * , 203601 ( 2008 ) ] that above threshold a tem@xmath0 mode following a random rotation in the transverse plane emerges in this system ( we denote it as _ bright _ mode ) , breaking thus its rotational invariance .
then , owed to the undetermination of the mode orientation , we showed that the phase quadrature of the transverse mode orthogonal to this one ( denoted as _ dark _ mode ) is perfectly squeezed at any pump level and without an increasing of the fluctuations on its amplitude quadrature ( which seems to contradict the uncertainty principle ) . in this article
we go further in the study of this system and analyze some important features not considered previously .
first we show that the apparent violation of the uncertainty principle is just that , apparent , as the conjugate pair of the squeezed quadrature is not another quadrature but the orientation of the bright mode ( which is completely undetermined in the long term ) .
we also study an homodyne scheme in which the local oscillator is not perfectly matched to the dark mode , as this could be impossible in real experiments due to the random rotation of the mode , showing that even in this case large levels of noise reduction can be obtained ( also including the experimentally unavoidable phase fluctuations ) .
finally we show that neither the adiabatic elimination of the pump variables nor the linearization of the quantum equations are responsible for the remarkable properties of the dark mode ( things which we prove analytically and through numerical simulations respectively ) , which were simplifying assumptions used in [ phys .
rev .
lett .
* 100 * , 203601 ( 2008 ) ] .
these studies show that the production of non - critically squeezed light through the spontaneous rotational symmetry breaking is a robust phenomenon . |
in this paper , we use symplectic reduction to derive the equations of motion for a rigid body moving in a two - dimensional fluid with point vortices . despite the fact that this setup is easily described
, it may come as a surprise that these equations were derived ( using straightforward calculations ) only recently by @xcite and @xcite . in order to see why this is so , despite the long history of this kind of problems , and in order to set the stage for our approach ,
let us trace some of the history of fluid - rigid body interaction problems .
[ [ rigid - bodies - in - potential - flow . ] ] rigid bodies in potential flow .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the equations of motion for a rigid body of mass @xmath0 and inertia tensor @xmath1 were first described by @xcite and are given by @xmath2 where @xmath3 and @xmath4 are the linear and angular velocity of the body , while @xmath5 and @xmath6 are the linear and angular momentum , related by @xmath7 and @xmath8 . here
, @xmath9 is the total mass of the rigid body , consisting of the mass @xmath0 and the _ virtual mass _ @xmath10 induced by the fluid .
similarly , @xmath11 , where @xmath12 is the virtual inertia tensor due to the fluid . the main difference between these equations and the euler equations governing the motion of a rigid body in vacuum is the appearance of the non - zero term @xmath13 on the right - hand side of the equation for @xmath5 . in other words ,
the center of mass no longer describes a uniform straight trajectory and is a non - trivial degree of freedom . from a geometric point of view , the motion of a rigid body in a potential flow can therefore be considered as a curve on the special euclidian group @xmath14 consisting of all translations and rotations in @xmath15 , where the former describe the orientation of the body while the latter encode the location of the center of mass . the kinetic energy for the rigid body in a potential flow is a quadratic function and hence determines a metric on @xmath14 .
the physical motions of the rigid body are geodesics with respect to this metric . by noting that the dynamics is invariant under the action of @xmath14 on itself , the system can be reduced from the phase space @xmath16 down to one on the dual lie algebra @xmath17 : in this way
, one derives the kirchhoff equations ( [ kirchhoffeq ] ) .
the situation is similar for a planar rigid body moving in a two - dimensional flow : the motion is then a geodesic flow on @xmath18 , the euclidian group of the plane , and reduces to a dynamical system on @xmath19 .
the dynamics of the reduced system is still described by the kirchhoff equations , but additional simplifications occur : since both @xmath6 and @xmath4 are directed along the @xmath20-axis , the right - hand side of the equation for @xmath6 is zero . throughout this paper , we will deal with the case of planar bodies and two - dimensional flows only .
the procedure of reducing a mechanical system on a lie group @xmath21 whose dynamics are invariant under the action of @xmath21 on itself is known as * _ lie - poisson reduction_*. it was pointed out by @xcite that both the euler equations for a rigid body as well as the euler equations for a perfect fluid can be derived using this approach . for more about the history of these and related reduction procedures , we refer to @xcite .
the geometric outlook on the kirchhoff equations , developed by @xcite , turned out to be crucial in the study stability results for bottom - heavy underwater vehicles ; see also @xcite .
[ [ point - vortices . ] ] point vortices .
+ + + + + + + + + + + + + + + another development in fluid dynamics to which the name of kirchhoff is associated , is the motion of point vortices in an inviscid flow
. a * _ point vortex _ * is defined as a singularity in the vorticity field of a two - dimensional flow : @xmath22 , where the constant @xmath23 is referred to as the * _ strength _ * of the point vortex . by plugging a superposition of @xmath24 point vortices into the euler equations for a perfect fluid , one obtains the following set of odes for the evolution of the vortex locations @xmath25 , @xmath26 : @xmath27 where @xmath28 and @xmath29 is the so - called * _ kirchhoff - routh function _ * , which is derived using the green s function for the laplacian with appropriate boundary conditions .
we will return to the explicit form for @xmath29 later .
for point vortices moving in an unbounded domain , @xmath29 was derived by routh and kirchhoff , whereas @xcite studied the case of vortices in a bounded container with fixed boundaries .
it is useful to recall here how the point vortex system is related to the dynamics of an inviscid fluid . in order to do so , we recall the deep insight of @xcite , who recognized that the motion of a perfect fluid in a container @xmath30 is a geodesic on the group @xmath31 of volume - preserving diffeomorphisms of @xmath30 , in much the same way as the motion of a rigid body is a geodesic on the rotation group @xmath32 .
the group @xmath31 acts on itself from the right and leaves the fluid kinetic energy invariant , a result known as * _ particle relabelling symmetry_*. hence , by lie - poisson reduction , the system can be reduced to one on the dual @xmath33 of the lie algebra of @xmath31 , and the resulting equations are precisely euler s equations .
moreover , any group acts on its dual lie algebra through the co - adjoint action , and a trajectory of the reduced mechanical system starting on one particular orbit of that action is constrained to remain on that particular orbit .
@xcite showed that for a perfect fluid , the co - adjoint orbits are labelled by vorticity and , when specified to the co - adjoint orbit corresponding to the vorticity of @xmath24 point vortices , euler s equations become precisely the vortex equations ( [ vortexeq ] ) . for a detailed overview of the geometric approach to fluid dynamics
, we refer to @xcite .
the dynamics of point vortices is treated by different authors , most notable @xcite , @xcite and @xcite .
[ [ the - rigid - body - interacting - with - point - vortices . ] ] the rigid body interacting with point vortices .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + given the interest in both rigid bodies and point vortices , it is natural to study the dynamics of a planar rigid body interacting with @xmath24 point vortices .
surprisingly , it was nt until the recent work of @xcite ( smbk ) and @xcite ( bmr ) that the equations of motion for this dynamical system were established .
both groups proceeded through an ad - hoc calculation to derive the equations of motion , and showed that the resulting equations are in fact hamiltonian .
however , both sets of equations are different at first sight . the smbk equations are formally identical to the kirchhoff equations ( [ kirchhoffeq ] ) together with the point vortex equations ( [ vortexeq ] ) , but the definitions of the momenta @xmath34 and the hamiltonian @xmath35 are modified to include the effect of the rigid body ( through the ambient fluid ) on the point vortices , and vice versa .
the configuration space is @xmath36 and the poisson structure is the sum of the poisson structures on the individual factors . from the bmr point of view , the dynamic variables are the velocity @xmath3 and angular velocity @xmath6 , together with the locations of the vortices . the hamiltonian is simply the sum of the kinetic energies for both subsystems , and to account for the interaction between the point vortices and the rigid body , bmr introduce a non - standard poisson structure involving the stream functions of the fluid .
somewhat miraculously , the equations of motion obtained by smbk and bmr turn out to be equivalent : @xcite establishes the existence of a poisson map taking the canonical smbk poisson structure into the bmr poisson structure .
however , a number of questions therefore remain .
most importantly , it is not obvious why one mechanical problem should be governed by two poisson structures , which are at first sight very different but turn out to be related by a certain poisson map .
moreover , it is entirely non - obvious why this system is hamiltonian in the first place : in the work of smbk and bmr , the hamiltonian structure is derived only afterwards by direct inspection .
[ [ main - contributions - and - outline - of - this - paper . ] ] main contributions and outline of this paper .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we shed more light on the issues addressed above by uncovering the geometric structures that govern this problem .
we begin this paper by giving an overview of rigid body dynamics and aspects of fluid mechanics in section [ sec : fluidsolid ] .
this material is mostly well - known and serves to set the tone for the rest of the paper . in section
[ sec : neumann ] we then introduce the so - called neumann connection , giving the response of the fluid to an infinitesimal motion of the rigid body . this connection has been described before , but a detailed overview of its properties seems to be lacking . in particular ,
we derive an expression for the curvature of this connection when the fluid space is an arbitrary riemannian manifold , generalizing a result of @xcite .
the remainder of the paper is then devoted to using reduction theory to obtain the equations of motion for the fluid - solid system in a systematic way .
we will derive the equations of motion for this system by reformulating it first as a geodesic flow on the cartesian product of the group @xmath37 of translations and rotations in 2d , and a space of embeddings describing the fluid configurations .
two symmetry groups act on this space : the particle relabelling symmetry group @xmath31 , and the group @xmath37 itself .
dividing out by the combined action of these symmetries is therefore an example of * _ reduction by stages_*. point vortices are the particle relabelling group @xmath31 and the euclidian group @xmath18 . ] after symplectic reduction with respect to @xmath31 in section [ sec : diff ] , we obtain a dynamical system on the product space @xmath38 , where the first factor is the phase space of the rigid body , while the second factor describes the locations of the point vortices .
the dynamics is governed by a * _ magnetic symplectic form _ * : it is the sum of the canonical symplectic forms on both factors , together with a non - canonical interaction term . from a physical point of view
, the interaction term embodies the interaction between the point vortices and the rigid body through the ambient fluid .
mathematically speaking , the map associating to each rigid body motion the corresponding motion of the fluid can be viewed as a connection , and the magnetic term is closely related to its curvature .
note that it is essential here to do _
reduction with respect to the particle relabeling symmetry rather than poisson reduction , even though the latter might be conceptually simpler .
recall that symplectic reduction reduces the system to a co - adjoint orbit ; as we shall see below , in the case of fluids interacting with solids these co - adjoint orbits are precisely labelled by the vorticity of the fluid .
symplectic reduction in particular the selection of one particular level set of the momentum map therefore amounts to imposing a specific form for the vorticity field . in our case , this is where we introduce the assumption that the vorticity is concentrated in @xmath24 point vortices . in the case of poisson reduction
, we would have obtained the equations of motion for a rigid body interacting with an arbitrary vorticity field .
after factoring out the particle relabeling symmetry , the resulting dynamical system is invariant under translations and rotations in the plane and can then be reduced with respect to the group @xmath37 .
this is the subject of section [ sec : redse ] .
physically , this corresponds to rewriting the equations of motion obtained after the first reduction in body coordinates .
however , because of the presence of the magnetic term in the symplectic form and the fact that @xmath37 acts diagonally on @xmath38 , this is not a straightforward task .
first , we derive the reduced poisson structure on the reduced space @xmath36 . because of the magnetic contributions to the symplectic form , the reduced poisson structure is not just the sum of the poisson structures on the individual factors , but includes certain non - canonical contributions as well .
we then show that the momentum map for the @xmath18-symmetry naturally defines a poisson map taking this poisson structure into the product poisson structure , possibly with a cocycle if the momentum map is not equivariant ( this happens when the total strength of the point vortices is nonzero ) .
we do the computations for a general product @xmath39 first , where @xmath35 is a lie group , @xmath40 is a symplectic manifold , and the product is equipped with a magnetic symplectic form . in this way , we generalize the `` coupling to a lie group '' scenario ( see @xcite ) to the case where magnetic terms are present . in this way
, the results of smbk and bmr are put on a firm geometric footing : the bmr poisson structure is the one obtained through reduction and involves the interaction terms , while the smbk poisson structure is simply the product poisson structure . the poisson map induced by the @xmath37-momentum map described above turns out to be precisely shashikanth s poisson map . as a consequence
, we also obtain an explicit prescription for the symplectic leaves of this system .
[ [ relation - with - other - approaches . ] ] relation with other approaches .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our method consists of rederiving the smbk and bmr equations by reformulating the motion of a rigid body in a fluid as a geodesic problem on the space @xmath41 . by imposing the assumption that the vorticity is concentrated in @xmath24 point vortices , and dividing out the symmetry , we obtain first of all the bmr equations , and secondly ( after doing a momentum shift ) the smbk equations .
this procedure is worked out in the body of the paper here we would like to reflect on the similarities with other dynamical systems .
recall that the dynamics of a particle of charge @xmath42 in a magnetic field @xmath43 can be described in two ways .
one is by using canonical variables @xmath44 and the hamiltonian @xmath45 while for the other one we use the kinetic energy hamiltonian @xmath46 but now we modify the symplectic form to be @xmath47 a simple calculation shows that both systems ultimately give rise to the familiar lorentz equations . from a mathematical point of view , this can be seen by noting that the * _ momentum shift map _ * @xmath48 maps the dynamics of the former system into that of the latter : @xmath49 in other words , both formulations are related by a symplectic isomorphism , thus making them equivalent .
an over - arching way of looking at the dynamics of a charged particle in a magnetic field is as a geodesic problem ( with respect to a certain metric ) on a higher - dimensional space : this is part of the famed * _ kaluza - klein approach_*. in this case , spacetime is replaced by the product manifold @xmath50 and charged particles trace out geodesics on this augmented space .
the standard , four - dimensional formulation of the dynamics can then be obtained by dividing out by the internal @xmath51-symmetry , resulting in the familiar lorentz equations on @xmath52 .
more information about these constructions can be found in @xcite , @xcite and @xcite .
our approach to the fluid - structure problem is similar but more involved because of the presence of two different symmetry groups .
however , the underlying philosophy is the same : by reformulating the motion of a rigid body and the fluid as a geodesic problem on an infinite - dimensional manifold , we follow the philosophy of kaluza - klein of trading in the complexities in the equations of motion for an increase in the number of dimensions of the configuration space
. then , by dividing out the @xmath53-symmetry , we obtain a reduced dynamical system governed by a magnetic symplectic form ( the analogue of ( [ magsymplectic ] ) ) , which is mapped after a suitable momentum shift @xmath54 into an equivalent dynamical system governed by the canonical symplectic form but with a modified hamiltonian . after reducing by the residual @xmath18-symmetry ,
the former gives rise to the bmr bracket , while the latter is nothing but the smbk system .
[ [ a - note - on - integrability . ] ] a note on integrability .
+ + + + + + + + + + + + + + + + + + + + + + + + the case of a circular cylinder is distinguished because of the existence of an additional conservation law , associated with the symmetry that rotates the cylinder around its axis .
when the circular cylinder interacts with one external point vortex , a simple count of dimensions and first integrals suggests that this problem is integrable , a fact first proved by @xcite .
indeed , for a single vortex the phase space is a symplectic leaf of @xmath55 , which is generically four - dimensional . on the other hand , for the circular cylinder
there are two conservation laws : the total energy and the material symmetry associated with rotations around the axis of the cylinder , which hints at liouville integrability . in the case of an ellipsoidal cylinder with nonzero eccentricity , @xcite
gather numerical evidence to show that the interaction with one vortex is chaotic .
this is to be contrasted with the motion of point vortices in an unbounded domain ( see @xcite ) , which is integrable for three vortices or less .
[ [ acknowledgements . ] ] acknowledgements .
+ + + + + + + + + + + + + + + + + it is a pleasure to thank richard montgomery , paul newton , tudor ratiu and banavara shashikanth , as well as frans cantrijn , scott kelly , bavo langerock , jim radford , and clancy rowley , for useful suggestions and interesting discussions . j. vankerschaver is a postdoctoral fellow from the research foundation flanders ( fwo - vlaanderen ) and a fulbright research scholar , and wishes to thank both agencies for their support .
additional financial support from the fonds professor wuytack is gratefully acknowledged .
e. kanso s work is partially supported by nsf through the award cmmi 06 - 44925 .
j. e. marsden is partially supported by nsf grant dms-0505711 .
this section is subdivided into two parts . in the first , and longest , part we describe the general setup for a planar rigid body interacting with a 2d flow .
the second part is then devoted to discussing a number of simplifying assumptions that will make the subsequent developments clearer . by separating these assumptions from the main problem setting , we hope to convince the reader that the method outlined in this paper does not depend on any specific assumptions on the rigid body or the fluid , and can be generalized to more complex problems . at the end of this paper , we discuss how these assumptions can be relaxed .
[ [ kinematics - of - a - rigid - body . ] ] kinematics of a rigid body .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + throughout this paper , we consider the motion of a * _ planar rigid body _ * interacting with a * _ 2d flow_*. we introduce an inertial frame @xmath56 , where @xmath57 span the plane of motion and @xmath58 is perpendicular to it .
the configuration of the rigid body is then described by a rotation with angle @xmath59 around @xmath58 and a vector @xmath60 describing the location of a fixed point of the body , which we take to be the center of mass . the orientation and position @xmath61 determine an element @xmath62 of the euclidian group @xmath37 given by @xmath63 written in this way
, the group composition and inversion in @xmath37 are given by matrix multiplication and inversion .
the lie algebra of @xmath37 is denoted by @xmath64 and essentially consists of infinitesimal rotations and translations .
its elements are matrices of the form @xmath65 it follows that @xmath64 can be identified with @xmath15 by mapping such a matrix to the triple @xmath66 .
we define a basis of @xmath64 given by @xmath67 for future reference , we also introduce a moving frame fixed to the rigid body , denoted by @xmath68 .
the transformation from body to inertial frame is given by @xmath69 , where @xmath70 and @xmath71 .
the * _ angular _ * and * _ translational velocities _ * of the rigid body relative to the inertial frame are defined as @xmath72 where dots denote derivatives with respect to time .
these quantities can be expressed in the body frame : the body angular velocity @xmath73 and the body translational velocity @xmath3 are related to the corresponding inertial quantities by @xmath74 for the case of a planar rigid body , the angular velocity is oriented along the axis perpendicular to the plane and so determines a scalar quantity : @xmath75 and @xmath76 . from a group - theoretic point of view , if the motion of a rigid body is given by a curve @xmath77 in @xmath37 , then we may define an element @xmath78 by putting @xmath79 and it can easily be checked that @xmath80 coincides with the body angular and translational velocities @xmath81 . the * _ kinetic energy _ * of the rigid body is given by the following expression : @xmath82 where @xmath83 is the moment of inertia of the body and @xmath84 is its mass .
the kinetic energy defines an inner product on @xmath64 , given by @xmath85 for @xmath86 and @xmath87 in @xmath64 , and where @xmath88 is given by @xmath89 here , @xmath90 is the @xmath91-by-@xmath91 identity matrix . by
left extension , this inner product induces a left - invariant metric on the whole of @xmath37 : @xmath92 for @xmath93 and @xmath94 in @xmath95 .
[ [ incompressible - fluid - dynamics . ] ] incompressible fluid dynamics .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the geometric description of an incompressible fluid goes back at least to the work of @xcite , who described the motion of an incompressible fluid in a fixed container @xmath30 as a geodesic on the diffeomorphism group of @xmath30 . in the case of a fluid interacting with a rigid body , the fluid container may change over time , reflecting the fact that the rigid body moves .
arnold s formulation can be extended to cover this case by considering the * _ space of embeddings _ * of the reference configuration of the fluid ( denoted by @xmath96 ) into @xmath97 .
recall that an embedding @xmath98 maps each reference point @xmath99 to its current configuration @xmath100 . in order to reflect the fact that the fluid is taken to be incompressible , we require that any fluid embedding @xmath101 is * _ volume preserving _ * :
if @xmath102 is the euclidian area element on @xmath97 and @xmath103 is a fixed area element on @xmath96 , then we require @xmath104 the space of all such volume - preserving embeddings is denoted by @xmath105 . in the sequel , we will specify additional boundary conditions on the fluid configurations , stating for example that the fluid is free to slide along the boundary of the solid , but this does not make any difference for the current expository treatment . a * _ motion _ * of a fluid is described by a curve @xmath106 in @xmath105 .
the * _ material velocity field _ * is the tangent vector field @xmath107 along the curve . here , @xmath108 is a map from @xmath96 to @xmath109 , whose value at a point @xmath110 is given by @xmath111 note that @xmath108 is not a vector field in the traditional sense ; rather , it is a vector field along the map @xmath112 .
in contrast , the * _ spatial velocity field _
* @xmath113 , defined as @xmath114 is a proper vector field , defined on @xmath115 .
the motion of a fluid can be described using the kinetic - energy lagrangian : @xmath116 where @xmath117 is the eulerian velocity field ( [ eulervelocity ] ) and the integration domain is the spatial domain of the fluid at time @xmath118 : @xmath119 .
just as in the case of the rigid body , this kinetic energy induces a metric on the space @xmath120 , given by @xmath121 by changing variables , this metric can be rewritten in spatial form as follows : @xmath122 where @xmath123 is the eulerian velocity associated to @xmath124 : @xmath125 , for @xmath126 , and the integration is again over the spatial domain @xmath127 of the fluid .
[ [ the - configuration - space - of - the - fluid - solid - system . ] ] the configuration space of the fluid - solid system .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the motion of a rigid body in an incompressible fluid combines aspects of both rigid - body and fluid dynamics .
we assume that the body occupies a circular region @xmath128 in the reference configuration , and that the remainder of the domain , denoted by @xmath96 , is taken up by the fluid .
the configuration space for the fluid - solid system is made up of pairs @xmath129 satisfying the following conditions . 1 .
the embedding @xmath101 represents the configuration of the fluid . in particular
, @xmath101 is volume - preserving , _ i.e. _ ( [ volforms ] ) is satisfied .
in addition , we assume that @xmath101 approaches the identity at infinity suitably fast .
2 . the element @xmath130 describes the configuration of the rigid body .
3 . the fluid satisfies a `` slip '' boundary condition : the normal velocity of the fluid coincides with the normal velocity of the solid , while the tangent velocity can be arbitrary , reflecting the fact that there is no viscosity in the fluid .
mathematically , this boundary condition is imposed by requiring that , as sets , @xmath131 is equal to @xmath132 , where @xmath130 is interpreted as a linear embedding of @xmath128 into @xmath97 .
we denote the space of all such pairs as @xmath41 ; this is a submanifold of @xmath133 .
the kinetic energy of the fluid - solid system is given by the sum of the rigid - body energy @xmath134 and the kinetic energy @xmath135 of the fluid : @xmath136 similarly , there exists a metric on @xmath41 given by the sum of the metrics ( [ metric : se ] ) and ( [ embmetric ] ) : @xmath137 the dynamics of rigid bodies moving in perfect fluids was studied before by @xcite , @xcite , and @xcite .
a similar configuration space , but with the @xmath37-factor replaced by a suitable set of smooth manifolds , was studied by @xcite for the dynamics of a liquid drop . [
[ particle - relabelling - symmetry . ] ] particle relabelling symmetry .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the kinetic energy @xmath135 of the fluid is invariant if we replace @xmath138 by @xmath139 , where @xmath140 is a volume - preserving diffeomorphism from @xmath96 to itself .
this represents the * _ particle relabelling symmetry _ * referred to in the introduction .
recall that a diffeomorphism @xmath141 is volume - preserving if @xmath142 , where @xmath103 is the volume element on @xmath96 .
the group of all volume - preserving diffeomorphisms is denoted by @xmath53 .
this group acts on the right on @xmath105 by putting @xmath143 , and hence also on @xmath41 .
the action of @xmath53 on @xmath41 makes @xmath41 into the total space of a principal fiber bundle over @xmath37 . in other words , if we define the projection @xmath144 as being the projection onto the first factor : @xmath145 , then the fibers of @xmath146 coincide precisely with the orbits of @xmath53 in @xmath41 .
[ [ vorticity - and - circulation . ] ] vorticity and circulation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + in classical fluid dynamics , the * _ vorticity _ * @xmath147 is defined as the curl of the velocity field : @xmath148 , and the * _ circulation _ * around the rigid body is the line integral of @xmath117 along any curve encircling the rigid body . in two dimensions , @xmath147 can be written as @xmath149 , where @xmath150 is called the scalar vorticity .
according to noether s theorem , there is a conserved quantity associated to the particle relabelling symmetry .
this conserved quantity turns out to be precisely the circulation of the fluid : @xmath151 and noether s theorem hence becomes kelvin s theorem , which states that circulation is materially constant . as a consequence of green s theorem , the circulation of the fluid is related to the vorticity : @xmath152 where @xmath153 is any surface whose boundary is @xmath154 .
it would lead us too far to explore the geometry of vorticity and circulation in detail ( for this , we refer to @xcite ) but at this stage we just note that _ the conservation of vorticity is closely linked to the particle relabelling symmetry_. [ [ the - helmholtz - hodge - decomposition . ] ] the helmholtz - hodge decomposition .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + if ( in addition to being incompressible ) the fluid is irrotational , meaning that @xmath155 , then there exists a velocity potential @xmath156 such that @xmath157 . in the presence of point vortices ,
the fluid is not irrotational but the velocity field @xmath117 can be uniquely decomposed in an irrotational part and a vector field @xmath158 representing the `` rotational '' contributions ( see for instance @xcite or @xcite ) .
this is the well - known helmholtz - hodge decomposition : @xmath159 here , @xmath158 is a divergence - free vector field which is tangent to the boundary of @xmath30 , while the potential @xmath156 is the solution to laplace s equation @xmath160 , subject to the boundary conditions that the normal derivative of @xmath156 equals the normal velocity of the rigid body , and that the velocity vanishes at infinity . in other words , one has @xmath161 and @xmath156 goes to zero as @xmath162 goes to infinity . here
@xmath163 and @xmath164 are the angular and translational velocity of the rigid body , expressed in a spatial frame , while @xmath165 represents the location of the center of mass .
since @xmath156 depends on @xmath166 and @xmath164 in a linear way , we may decompose @xmath156 , following kirchhoff , as @xmath167 here @xmath168 , @xmath169 , and @xmath170 are elementary velocity potentials corresponding to infinitesimal translations in the @xmath171- and @xmath172-direction and to a rotation , respectively .
they satisfy the laplace equation with the following boundary conditions : @xmath173 note that @xmath156 depends on the boundary data , and hence on @xmath174 , while the elementary potentials depend on the location of the rigid body ( encoded by @xmath62 ) only . for the sake of clarity
, we will suppress these arguments when no confusion is possible .
the elementary velocity potentials for a circular body are given below : the reader can check that they indeed depend on the location of the rigid body .
the helmholtz - hodge decomposition defines a connection on the principal fiber bundle @xmath175 . to see this , note that any tangent vector to @xmath41 is of the form @xmath176 .
we define its horizontal and vertical part by applying the helmholtz - hodge decomposition to @xmath177 , and we put @xmath178 we will verify in section [ sec : neumann ] below that this prescription indeed defines a connection , which we term the * _ neumann connection _ * , since its horizontal subspaces are found by solving the neumann problem ( [ neumann ] ) .
this connection was used in a variety of contexts , ranging from the dynamics of fluid drops ( see @xcite ) to problems in optimal transport ( see @xcite ) .
[ [ the - lie - algebra - of - divergence - free - vector - fields - and - its - dual . ] ] the lie algebra of divergence - free vector fields and its dual .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + at least on a formal level , @xmath53 is a lie group with associated to it a lie algebra , denoted by @xmath179 and consisting of divergence - free vector fields which are parallel to the boundary of @xmath96 .
the bracket on @xmath179 is the jacobi - lie bracket of vector fields ( which is the negative of the usual bracket of vector fields ) and its dual , denoted by @xmath180 , is the set of linear functionals on @xmath179 .
this set can be identified with the set of one - forms on @xmath96 modulo exact forms : @xmath181 ( see @xcite ) while the duality pairing between elements of @xmath179 and @xmath180 is given by @xmath182 \right > = \int_{\mathcal{f}_0 } \alpha(\mathbf{u } ) \ , \eta_0,\ ] ] where @xmath183 and @xmath184 \in \mathfrak{x}^{\ast } _ { \mathrm{vol}}$ ] .
note that the right - hand side does not depend on the choice of representative @xmath185 .
another interpretation of the dual lie algebra @xmath180 is as the set of exact two - forms on @xmath96 .
any class @xmath184 $ ] in @xmath186 is uniquely determined by the exterior differential @xmath187 and by the value of @xmath185 on the generators of the first homology of @xmath96 . in our case ,
the first homology group of @xmath96 is generated by any closed curve @xmath154 encircling the rigid body , and its pairing with @xmath185 is given by @xmath188 under this identification , @xmath187 represents the * _ vorticity _ * , while @xmath23 represents the * _ circulation_*. since in our case the circulation is assumed to be zero , it follows that @xmath184 $ ] is completely determined by @xmath189 . from a geometric point of view
, the vorticity field can be interpreted as an element of @xmath186 : if we assume that the fluid is moving on an arbitrary riemannian manifold , then the vorticity can be defined by @xmath190 here @xmath191 is the flat operator associated to the metric . in the case of euclidian spaces , this definition reduces to the one involving the curl of @xmath117 .
this definition is slightly different from the one in @xcite , where vorticity is defined as a two - form on the inertial space @xmath97 , whereas in our interpretation , vorticity lives on the material space @xmath96 . both definitions are related by push - forward and pull - back by @xmath101 , and hence carry the same amount of information .
our definition has the advantage that @xmath150 is naturally an element of @xmath186 , which is preferable from a geometric point of view .
we finish this section by noting that any lie group acts on its lie algebra and its dual lie algebra through the adjoint and the co - adjoint action , respectively . for the group of volume - preserving diffeomorphisms ,
both are given by pull - back : if @xmath140 is an element of @xmath53 , and @xmath117 and @xmath187 are elements of @xmath179 and @xmath186 , respectively , then @xmath192 in this section , we impose some specific assumptions on the rigid body and the fluid .
it should be pointed out that while these assumptions greatly simplify the exposition , the general reduction procedure can be carried out under far less stringent assumptions .
later on , we will discuss how some of these assumptions may be removed .
[ [ the - rigid - body . ] ] the rigid body .
+ + + + + + + + + + + + + + + in order to tie in this work with previous research efforts , we assume the rigid body to be * _ circular _ * with radius @xmath193 and neutrally buoyant ( _ i.e. _ its body weight @xmath84 is balanced by the force of buoyancy ) .
if the density of the fluid is set to @xmath194 , this implies that the body has mass @xmath195 .
the moment of inertia of the body around the axis of symmetry is denoted by @xmath83 . for a rigid planar body of circular shape , the elementary velocity potentials @xmath168
, @xmath169 , and @xmath170 occurring in ( [ kirchhoff ] ) can be calculated analytically ( see @xcite ) and are given by @xmath196 while @xmath197 , reflecting the rotational symmetry of the body .
here @xmath198 are the coordinates of the center of the disc . in some cases , it will be more convenient to express @xmath156 in body coordinates . in analogy with ( [ kirchhoff ] ) , we may write @xmath199 where @xmath200 and @xmath201 are the translational and angular velocity in the body frame , respectively . for the circular cylinder ,
the elementary potentials in body frame are given by @xmath202 note that @xmath203 and @xmath204 do not depend on the location of the rigid body , in contrast to @xmath205 and @xmath170 .
[ [ point - vortices.-1 ] ] point vortices .
+ + + + + + + + + + + + + + + as for the fluid , we make the fundamental assumption that the vorticity is concentrated in @xmath24 * _ point vortices _ * of strengths @xmath206 , @xmath207 , and that there is no circulation .
considered as a two - form on @xmath96 , the former means that the vorticity is given by @xmath208 where @xmath209 are coordinates on @xmath96 , and @xmath210 is the reference location of the @xmath211th vortex , @xmath212 . as pointed out above , @xmath150 is an element of @xmath186 .
[ [ the - kirchhoff - routh - function . ] ] the kirchhoff - routh function .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xcite showed that the kinetic energy for the vortex system is the negative of the kirchhoff - routh function @xmath29 for a system of @xmath24 point vortices moving in a domain with moving boundaries : @xmath213 the precise form of @xmath29 is given by @xmath214 where @xmath215 is a green s function for the laplace operator of the form @xmath216 the function @xmath62 is harmonic in the fluid domain and is the stream function of @xmath217 .
as pointed out in @xcite , in the case of a circular cylinder , @xmath62 can be calculated explicitly using milne - thomson s circle theorem ( @xcite ) , and is given by @xmath218 from a geometrical point of view , @xcite showed that the vortex energy can be obtained ( up to some `` self - energy '' terms ) by inverting the relation @xmath219 , where @xmath150 is the scalar part of ( [ vorticity ] ) , and substituting the resulting velocity field @xmath117 generated by @xmath24 vortices into the expression for the kinetic energy of the fluid .
although the analysis of @xcite was for an unbounded fluid domain , their result can easily be extended to the case considered here , yielding again the negative of the kirchhoff - routh function as in ( [ kr ] ) .
[ [ dynamics - of - the - fluid - solid - system . ] ] dynamics of the fluid - solid system .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as stated in the introduction , @xcite ( smbk ) were the first to derive the equations of motion for a rigid cylinder interacting with @xmath24 point vortices .
these equations generalize both the kirchhoff equations for a rigid body in a potential flow and the equations for @xmath24 point vortices in a bounded flow . rather remarkably , smbk established by direct inspection that these equations are hamiltonian with respect to the canonical poisson structure on @xmath220 ( _ i.e. _ the sum of the poisson structures on both factors ) , and a hamiltonian given below involving the kinetic energy and interaction terms .
the smbk equations are given by @xmath221 where @xmath5 and @xmath222 are the translational and angular momenta of the system , defined by @xmath223 and @xmath35 is the hamiltonian : @xmath224 here , @xmath225 is the total mass of the cylinder , consisting of the intrinsic mass @xmath84 and the added mass @xmath226 ( due to the presence of the fluid ) : @xmath227 .
the smbk equations are hamiltonian on the space @xmath228 equipped with the poisson bracket @xmath229 here , @xmath230 and @xmath21 are functions on @xmath231 . in the theorem
above , @xmath232 is the lie - poisson bracket on @xmath233 : @xmath234 for arbitrary functions @xmath235 on @xmath233 .
similarly , @xmath236 is the vortex bracket , given by : @xmath237 where @xmath238 are arbitrary functions on @xmath239 .
[ [ the - bmr - equations . ] ] the bmr equations .
+ + + + + + + + + + + + + + + + + + a completely different perspective on the rigid body interacting with @xmath24 point vortices is offered by @xcite ( bmr ) . from their point of view
, the equations of motion are again written in hamiltonian form on @xmath36 , but now with a _
noncanonical _ poisson bracket : @xmath240 for all functions @xmath230 on @xmath36 .
the hamiltonian is the sum of the kinetic energies of the subsystems , without interaction terms : @xmath241 whereas the poisson bracket is determined by its value on the coordinate functions : @xmath242 where @xmath243 , @xmath244 , @xmath245 , and @xmath23 is the total vortex strength : @xmath246 note that this poisson bracket differs from the one in @xcite by an overall factor of @xmath247 .
@xcite showed that this discrepancy can be attributed to the way in which bmr choose the fluid density .
[ [ the - link - between - the - smbk - and - the - bmr - equations . ] ] the link between the smbk and the bmr equations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + by explicit calculation , @xcite constructed a poisson map taking the smbk equations into the bmr equations .
his result is listed below .
[ thm : shashi ] the map @xmath248 , where @xmath5 and @xmath249 are given by , is a poisson map from @xmath36 equipped with the bracket @xmath250 to @xmath36 with the bracket @xmath251 .
even though this result asserts that both sets of equations are equivalent , it leaves open the question as to _ why _ this is so .
by re - deriving the equations of motion using symplectic reduction , not only do we obtain both sets of equations , but the map
@xmath153 also follows naturally .
the bundle @xmath144 is equipped with a principal fibre bundle connection , called the * _ neumann connection _ * by @xcite .
there are many ways of describing this connection , but from a physical point of view , the definition using the horizontal lift operator is perhaps most appealing . from this point of view , the neumann connection is a map @xmath252 where @xmath156 is the solution of the neumann problem ( [ neumann ] ) associated to @xmath253 . in other words ,
the neumann connection associates to each motion @xmath253 the corresponding induced velocity field of the fluid , and hence encodes the effect of the body on the fluid .
it is important to note that the neumann connection does not depend on the point vortex model and is valid for any vorticity field .
similar connections as this one have been described before ( see for example @xcite ) but a complete overview of its definition and properties seems to be lacking . in this section
, we give an outline of the properties of the neumann connection which are relevant for the developments in this paper , leaving detailed proofs for the appendix .
[ [ invariance - of - the - kirchhoff - decomposition . ] ] invariance of the kirchhoff decomposition .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + before introducing the neumann connection , we prove that the velocity potential @xmath254 is left @xmath37-invariant , expressing the fact that the dynamics is invariant under translations and rotations of the combined solid - fluid system .
[ prop : invariance ] the velocity potential @xmath156 is left @xmath37-invariant in the sense that @xmath255 for all @xmath256 and @xmath257 .
this assertion can be proved in a number of different ways .
the easiest is to use the assumption that the body is circular and solve the equation for the elementary potentials explicitly . recall that these elementary potentials are given by ( [ elempot ] ) .
it is then straightforward to check that ( [ invariance ] ) holds , using the transformation properties of the velocity in the inertial frame .
[ [ the - connection - one - form . ] ] the connection one - form .
+ + + + + + + + + + + + + + + + + + + + + + + + for our purposes , it is convenient to define the neumann connection through its connection one - form @xmath258 given by @xmath259 where @xmath158 is the divergence - free part in the helmholtz - hodge decomposition ( [ hhdecomp ] ) of the eulerian velocity @xmath260 . a proof that this prescription determines a well - defined connection form can be found in the appendix , proposition [ prop : conn ] , where it is also shown that this prescription agrees with the horizontal lift operator ( [ holift ] ) .
[ [ the - curvature - of - the - neumann - connection . ] ] the curvature of the neumann connection .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + it will be convenient in what follows to have an expression for the @xmath150-component of the curvature of the neumann connection , where for now @xmath150 is an arbitrary element of @xmath186 .
later on , @xmath150 will be the vorticity ( [ vorticity ] ) associated to @xmath24 point vortices .
the curvature of a principal fiber bundle connection is a two - form @xmath261 whose definition is listed in the appendix . for the neumann connection we have @xmath262),\ ] ] where @xmath263 and @xmath264 are horizontal vector fields on @xmath41 such that @xmath265 and @xmath266 .
[ prop : curv ] let @xmath267 and @xmath268 be elements of @xmath269 and denote the solutions of the neumann problem ( [ neumann ] ) associated to @xmath93 resp .
@xmath270 by @xmath271 and @xmath272 .
then the @xmath150-component of the curvature @xmath261 is given by @xmath273 where @xmath274 is the metric on the space of forms on @xmath30 , induced by the euclidian metric on @xmath30 .
let @xmath150 be equal to @xmath275 , and pick @xmath185 such that @xmath276 , where @xmath158 is a divergence - free vector field on @xmath30 tangent to @xmath277 .
for example , in the case of the vorticity due to @xmath24 point vortices , @xmath158 is the velocity field due to the vortices and their images ( see @xcite ) .
the calculation of the curvature involves computing the jacobi - lie bracket of two horizontal vector fields and taking the divergence - free part of the result . because of the special form of @xmath185 we can dispense with the latter step , since
@xmath185 is chosen to be @xmath278-orthogonal to gradient vector fields .
therefore , the @xmath150-component of the curvature is given by @xmath279)\right > \\ & = - \int_{\mathcal{f}_0 } ( \varphi^{\ast } \alpha)([x^h , y^h ] ) \
, \eta_0 , \end{aligned}\ ] ] where the bracket on the left - hand side is the jacobi - lie bracket , which is the negative of the usual commutator of vector fields .
the bracket can be made more explicit by noting that ( as vector fields on @xmath41 ) @xmath280(g , \varphi ) = ( [ ( \mathbf{d } \phi_1)^\flat , ( \mathbf{d } \phi_2)^\flat ] \circ \varphi , \cdots),\ ] ] where the dots denote a term in @xmath281 whose explicit form does nt matter .
the curvature then becomes @xmath282 ) \ , d \mathbf{x}.\ ] ] the remainder of the proof relies on the following formula for the codifferential of a wedge product ( ( * ? ? ?
* formula 1.34 ) ) : @xmath283^\sharp,\ ] ] where @xmath284 . applying ( [ codiff ] ) with @xmath285 and @xmath286 gives @xmath287^\sharp = \delta \phi_1 \mathbf{d } \phi_2 - \delta \phi_2 \mathbf{d } \phi_1 - \delta(\mathbf{d } \phi_1 \wedge \mathbf{d } \phi_2),\ ] ] where @xmath288 is the laplace - beltrami operator on differential forms .
since both @xmath271 and @xmath272 are harmonic , the first two terms of the right - hand side vanish . consequently , the @xmath150-component of the curvature becomes @xmath289 using the adjointness of @xmath290 and @xmath291 for manifolds with boundary ( this is a simple consequence of stokes theorem ) . in traditional fluid mechanics notation , formula ( [ curvature ] )
becomes @xmath292 a formula first derived in @xcite .
as mentioned in the introduction , the fluid - solid system on the space @xmath293 is invariant under the particle relabelling group @xmath53 .
we now perform symplectic reduction to eliminate that symmetry . in order to do so
, we need to fix a value of the momentum map associated to the @xmath53-symmetry . from a physical point of view , this boils down to fixing the vorticity of the system ; it is at this point that the assumption is used that the vorticity is concentrated in @xmath24 point vortices . before tackling symplectic reduction in the context of the fluid - solid system
, we first give a general overview of cotangent bundle reduction following @xcite .
roughly speaking , applying symplectic reduction to a cotangent bundle yields a space which is diffeomorphic to the product of a reduced cotangent bundle and a co - adjoint orbit of the group .
the dynamics on the reduced space is governed by a reduced hamiltonian and a magnetic symplectic form : the reduced symplectic form is the sum of the canonical symplectic forms on the individual factors and an additional magnetic term . as we shall see below , in the case of the fluid - solid system ,
the reduced phase space is equal to @xmath38 , where the first factor describes the rigid body , while the second factor determines the configuration of the vortices . at first sight
, it may therefore appear that the intermediate fluid is completely gone .
yet , the vortices act on the rigid body , and vice versa , through the surrounding fluid .
the answer to this apparent contradiction is that the effect of the fluid is concentrated in the magnetic symplectic form , for which we derive a convenient expression below . in this section ,
we collect some relevant results from @xcite .
we consider a manifold @xmath41 on which a lie group @xmath21 , with lie algebra @xmath294 , acts from the right and we denote the action by @xmath295 .
in addition , we assume that we are given a connection one - form @xmath296 with curvature two - form @xmath261 . in the rest of this paper , @xmath41 will be the configuration space of the solid - fluid system , while the structure group @xmath21 will be @xmath53 , and the connection @xmath297 the neumann connection .
[ [ the - curvature - as - a - two - form - on - the - reduced - space . ] ] the curvature as a two - form on the reduced space .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath150 be an element of @xmath298 , and denote its isotropy subgroup under the co - adjoint action by @xmath299 : @xmath300 consider the contraction @xmath301 of @xmath150 with the curvature @xmath261 : due to the @xmath21-equivariance and the fact that @xmath261 vanishes on vertical vectors ( see ( [ props ] ) ) , we may show that @xmath301 is a @xmath299-invariant form .
the @xmath150-component of the curvature @xmath261 is a @xmath299-invariant two - form : 1 .
@xmath302 , for all @xmath303 ; 2 .
@xmath304 for all @xmath305 .
hence , @xmath301 induces a two - form @xmath306 on @xmath307 such that @xmath308 where @xmath309 is the quotient map . for the first property
, we have @xmath310 if @xmath303 . the second item follows from the corresponding property for @xmath261 . actually , more is true : @xmath311 for all @xmath312 , not just in @xmath313 .
[ [ cotangent - bundle - reduction . ] ] cotangent bundle reduction .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + the first stage in reducing the phase space consists of dividing out the particle relabelling symmetry . to this end
, we use the framework of * _ cotangent bundle reduction _ * to construct the reduced phase space .
the framework for cotangent bundle reduction outlined in @xcite allows us to write down the reduced phase space and the modified symplectic form .
we quote : [ thm : cotred ] ( marsden _ et al . _ [ theorem 2.2.1 ] ) 1 .
there exists a symplectic imbedding @xmath314 of the reduced phase space @xmath315 into the cotangent bundle @xmath316 with the shifted symplectic structure @xmath317 .
the image of @xmath314 is the vector subbundle @xmath318^\circ$ ] of @xmath319 , where @xmath320 is the vector subbundle consisting of vectors tangent to the @xmath21-orbits in @xmath41 , and @xmath321 denotes the annihilator relative to the natural duality pairing between @xmath322 and @xmath323 .
here @xmath315 is the reduced phase space @xmath324 , where @xmath299 is the isotropy subgroup of @xmath150 .
the two - form @xmath325 is called the * _ magnetic two - form _ * , and is defined as @xmath326 , where @xmath327 is the cotangent bundle projection and @xmath328 is a two - form on @xmath307 . to define @xmath328 , consider the one - form @xmath329 on @xmath41 .
one can show that @xmath330 ( but not @xmath331 itself ) is @xmath299-invariant , and the induced two - form on @xmath307 is precisely @xmath328 , or in other words @xmath332 where @xmath333 is the quotient map .
the calculation of @xmath330 is greatly facilitated by using the cartan structure equation ( @xcite ) , which allows us to rewrite @xmath334 as @xmath335 \right>.\ ] ] observe the sign difference on the right - hand side with @xcite , which is due to the fact that we take @xmath21 to be acting on @xmath41 from the _
right_. theorem [ thm : cotred ] provides us with an explicit prescription for the reduced phase space .
recall that the diffeomorphism group @xmath53 acts on the dual lie algebra @xmath336 through pull - back . in particular , if @xmath150 represents the vorticity due to @xmath24 point vortices as in ( [ vorticity ] ) , then @xmath337 in other words , the diffeomorphism group acts on the space of point vortices by simply moving the vortices .
it also follows that the isotropy subgroup of @xmath150 consists of all diffeomorphisms @xmath140 for which the vortex reference locations are fixed : @xmath338 , for @xmath26 .
we denote the group of all such diffeomorphisms as @xmath339 .
the group @xmath339 acts on @xmath41 and moreover , the quotient of @xmath41 by this action is diffeomorphic to @xmath340 . to see this , note that @xmath339 acts on the @xmath120 factor only , and that there exists a diffeomorphism of the quotient space @xmath341 with @xmath239 , given by @xmath342 \mapsto ( \varphi(\mathbf{\bar{x}}_1 ) , \ldots , \varphi(\mathbf{\bar{x}}_n)).\ ] ] similarly , the projection of @xmath41 onto the quotient space @xmath343 is given by @xmath344 after reducing by the group of volume preserving diffeomorphisms , the reduced phase is given by @xmath38 .
the vertical bundle @xmath345 on @xmath41 consists of vectors @xmath346 .
projecting this bundle down under @xmath347 shows that @xmath348 is spanned by elements of the form @xmath349 and as each of the @xmath24 vectors on the right hand side can range over the whole of @xmath109 , @xmath348 is equal to @xmath350 .
its annihilator is therefore @xmath351 .
the reduced symplectic structure on @xmath38 is described in theorem [ thm : cotred ] .
explicitly , it is given by @xmath352 where @xmath353 is the canonical symplectic structure on @xmath354 and @xmath325 is the pullback to @xmath355 of the form @xmath328 on @xmath340 .
from now on , we will no longer make any notational distinction between @xmath328 and its pull - back @xmath325 and denote both by @xmath328 .
apart from the reduced symplectic form , which will be determined explicitly later on , the dynamics on the reduced space is also governed by a reduced version of the hamiltonian ( [ totalenergy ] ) .
the calculation of this hamiltonian is the subject of the next section .
the kinetic energy ( [ kinenergy ] ) is invariant under the action of @xmath339 on @xmath41 and hence determines a reduced kinetic energy function on the quotient space @xmath340 .
parts of the computation of the explicit form of the reduced kinetic energy can be found throughout the literature , but since no single reference has a complete picture , we briefly recall these results here . using the helmholtz - hodge decomposition ( [ hhdecomp ] )
, the kinetic energy ( [ kinenergy ] ) of the combined solid - fluid system can be written as @xmath356 where we have used the fact that @xmath357 and @xmath358 are @xmath278-orthogonal . following a reasoning similar to @xcite
, it can be shown that the kinetic energy of the vortex system is nothing but the negative of the kirchhoff - routh function @xmath29 ( see ( [ krfunction ] ) ) : @xmath359 the gradient term in ( [ redkin ] ) can be rewritten using a standard procedure , going back to kirchhoff and lamb , and yields the familiar added masses and moments of inertia for a rigid body in potential flow .
by using green s theorem to rewrite the gradient term as an integral over the boundary , and substituting the kirchhoff expansion ( [ kirchhoff ] ) , one can show that ( see @xcite ) @xmath360 where @xmath361 is a @xmath362-by-@xmath362 matrix of added masses and inertia whose entries depend only on the geometry of the rigid body . for the circular cylinder , @xmath361 can be evaluated explicitly : @xmath363 the important point is to note that the gradient term ( [ gradterm ] ) has the same form as the kinetic energy of the rigid body . by introducing the matrix @xmath364 , where @xmath88 is the mass matrix ( [ mm ] ) , the kinetic energy of the rigid body , together with the gradient term ,
can be written as @xmath365 which determines by left extension a function , also denoted by @xmath366 , on @xmath95 . putting everything together ,
we conclude that the total reduced kinetic energy on @xmath367 is given by @xmath368 where @xmath29 is given by ( [ krfunction ] ) and ( [ green ] ) .
two issues are noteworthy here .
first of all , we follow @xcite and @xcite in `` regularizing '' @xmath29 by subtracting infinite contributions that arise when putting @xmath369 in the expression for the green s function @xmath21 in ( [ green ] ) . secondly , we have chosen to express the kinetic energy in the body frame , which will be more convenient later on .
recall that @xmath328 is a two - form on @xmath370 defined as @xmath371 where @xmath372 ( @xmath373 ) , and @xmath374 and @xmath375 are elements of @xmath281 .
the embedding @xmath101 has to satisfy @xmath376 for @xmath26 , while @xmath377 and @xmath378 should satisfy @xmath379 using cartan s structure equation ( [ cartan ] ) , @xmath380 can be rewritten as the difference of a curvature term and a `` lie - poisson '' term : if we introduce two - forms @xmath381 and @xmath382 on @xmath370 , determined by @xmath383\right>,\ ] ] then @xmath384 . in the computation of these terms
, we will frequently encounter expressions involving the vertical part of @xmath385 and @xmath386 , evaluated at the vortex locations .
we now derive a convenient expression for these terms .
let @xmath271 be the solution of the neumann problem ( [ neumann ] ) with boundary data @xmath374 and denote by @xmath117 the eulerian velocity field associated to @xmath377 . from the helmholtz - hodge decomposition @xmath387
, it follows the divergence - free part @xmath158 satisfies ( for @xmath26 ) @xmath388 since @xmath389 . from the definition of @xmath297
it then follows that we have proved the following fact about the vertical part of @xmath377 : @xmath390 for @xmath26 .
a similar property holds for @xmath378 , with @xmath391 replaced by @xmath392 and involving @xmath272 , the solution of ( [ neumann ] ) with boundary data @xmath375 . as an aside
, we note that the neumann connection induces a connection on the trivial bundle @xmath393 , and the expression between brackets on the right - hand side of ( [ vf ] ) is the vertical component of the vector @xmath391 .
this connection is important in the theory of _ routh reduction _ ( see @xcite ) .
[ [ the - curvature - term . ] ] the curvature term .
+ + + + + + + + + + + + + + + + + + + the curvature term @xmath381 is given by @xmath394 where @xmath261 is the curvature of the neumann connection .
it follows from ( [ curvature ] ) that @xmath395 does not depend on the value of @xmath391 and @xmath392 , but only on @xmath374 and @xmath396 . from now on
, we will therefore suppress these arguments .
for the rigid body interacting with @xmath24 point vortices , the curvature term is given by @xmath397 where @xmath271 and @xmath272 are the solutions of ( [ neumann ] ) associated to @xmath374 and @xmath375 , respectively , and @xmath398 is the volume element on @xmath30 .
we use the explicit form for the elementary velocity potentials derived in proposition [ prop : invariance ] .
note that @xmath395 depends on @xmath374 and @xmath375 through @xmath399 and @xmath400 , so we can calculate its expression on the basis @xmath401 of @xmath64 given in ( [ sebasis ] ) by assuming that @xmath402 and @xmath403 .
first , we prove that the boundary term in ( [ curvature ] ) is zero .
this is obvious if either @xmath374 or @xmath375 is proportional to @xmath404 , since @xmath197 .
we may therefore assume that @xmath405 and @xmath406 .
in this case , the two - form @xmath407 restricted to the boundary of @xmath30 ( the curve @xmath408 ) is given by @xmath409 .
since @xmath410 in two dimensions , the boundary term reduces to @xmath411 which is zero , since by assumption there is no circulation around the rigid body .
this leaves us with the integral over the fluid domain in ( [ curvature ] ) . plugging in the explicit form ( [ vorticity ] ) of the vorticity
, we have @xmath412 we do not work out this expression any further , since it will be cancelled out by a similar term in the computation of @xmath382 .
it is remarkable that the contribution from the curvature to the magnetic term will be cancelled in its entirety by a similar term in the expression for lie - poisson term , which is computed below .
however , in the case of non - zero circulation , the curvature generates an additional term ( [ curvcirc ] ) proportional to the circulation .
the effect of this term is studied in @xcite .
[ [ the - lie - poisson - term . ] ] the lie - poisson term .
+ + + + + + + + + + + + + + + + + + + + + the lie - poisson term @xmath382 is given by @xmath413 \right>,\end{gathered}\ ] ] where @xmath414 and @xmath415 have similar interpretations as before .
the right - hand side can be made more explicit by noting that , for divergence - free vector fields @xmath416 and @xmath417 tangent to @xmath277 and arbitrary one - forms @xmath185 , the following holds : @xmath418 ) \,d \textbf{x } = \int_\mathcal{f } \mathbf{d } \alpha(\mathbf{u}_1 , \mathbf{u}_2 ) \,d \textbf{x}.\ ] ] the proof of this assertion is a straightforward application of cartan s magic formula and parallels the proof of theorem 4.2 in @xcite . using this formula
, we have @xmath419 \right > = \int_\mathcal{f } ( \varphi^{\ast } \alpha)([\mathcal{a}_{(\varphi , g)}(\dot{\varphi}_1 , \dot{g}_1 ) , \mathcal{a}_{(\varphi , g)}(\dot{\varphi}_2 , \dot{g}_2 ) ] ) \ , d \textbf{x } \\ = \int_\mathcal{f }
\mu(\mathcal{a}_{(\varphi , g)}(\dot{\varphi}_1 , \dot{g}_1 ) , \mathcal{a}_{(\varphi , g)}(\dot{\varphi}_2 , \dot{g}_2 ) ) \ , d \textbf{x } \\ = \sum_{i=1}^n \gamma_i \ , ( \varphi^{\ast } d \textbf{x})\big((\mathcal{a}_{(\varphi , g)}(\dot{\varphi}_1 , \dot{g}_1))(\mathbf{\bar{x}}_i ) , ( \mathcal{a}_{(\varphi , g)}(\dot{\varphi}_2 , \dot{g}_2))(\mathbf{\bar{x}}_i)\big).\end{gathered}\ ] ] by substituting ( [ vf ] ) in this expression , we conclude that the lie - poisson term is given by @xmath420 [ [ the - magnetic - two - form - putting - everything - together . ] ] the magnetic two - form : putting everything together .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + using the previously derived results , we may conclude that @xmath328 is given by ( suppressing its arguments for the sake of clarity ) @xmath421 introducing stream functions @xmath422 ( @xmath423 ) as harmonic conjugates to @xmath424 , _
i.e. _ such that @xmath425 we may rewrite the expression for @xmath328 as @xmath426 the stream function can be written in lamb form by introducing the elementary stream functions @xmath427 , @xmath428 , and @xmath429 as harmonic conjugates to @xmath168 , @xmath169 , and @xmath170 , respectively . for @xmath126
, we then have @xmath430 and so the magnetic two - form becomes @xmath431 where the coordinates on @xmath37 are denoted by @xmath432 , and @xmath433 .
[ thm : magform ] the magnetic two - form @xmath328 is a two - form on @xmath434 and can be written as @xmath435 where @xmath436 is the one - form on @xmath437 given by @xmath438 note that the exterior derivative on the right - hand side of ( [ betaexpl ] ) is the exterior derivative on @xmath434 .
expanding the second term , we have @xmath439 where the subscript ` @xmath97 ' indicates that the derivative should be taken with respect to the spatial variables only , _
i.e. _ that the @xmath37-variable should be kept fixed . also , we rely on the fact that @xmath440 ( since @xmath197 ) , and that @xmath441 does not depend on @xmath59 .
the theorem is proved once we show that the term in brackets vanishes . to show this
, we need the explicit form for the elementary stream functions : @xmath442 and it is then easily shown that this is indeed the case . putting everything together ,
we conclude that the symplectic structure @xmath443 on @xmath444 ( see ( [ magsymp ] ) ) is given by @xmath445 where @xmath328 is given in the theorem above .
the first term is the canonical symplectic form on @xmath446 while the magnetic consists of two parts , one ( proportional to @xmath102 ) being the symplectic form on @xmath239 , while the other one encodes the effects of the ambient fluid through the stream functions @xmath447 and @xmath448 .
note that @xmath239 is a co - adjoint orbit of the diffeomorphism group , and that its symplectic structure is nothing but the konstant - kirillov - souriau form .
after reducing the fluid - solid system with respect to the @xmath53-symmetry , we are left with a system on @xmath38 with a magnetic symplectic form .
the group @xmath37 acts diagonally on the reduced phase space .
what remains is to carry out the reduction with respect to the group @xmath18 and then to make the connection with the equations of motion in @xcite and @xcite .
two factors complicate this strategy , however : first of all , there is the presence of the magnetic term in the symplectic structure , which precludes using standard euler - poincar theory , for instance . secondly , the phase space is a cartesian product on which the symmetry group acts diagonally . even in the absence of magnetic terms , this would lead to the appearance of certain * _ interaction terms _ * in the poisson structure ( see @xcite ) . in order to do poisson reduction for this kind of manifold
, we begin this section by considering the product of a cotangent bundle @xmath450 , where @xmath35 is a lie group , and a symplectic manifold @xmath40 , and we assume that the symplectic structure on @xmath451 is the sum of the canonical symplectic structures on both factors plus an additional magnetic term .
this case extends both magnetic lie - poisson reduction in @xcite , for which the @xmath40-factor is absent , as well as the `` coupling to a lie group '' scenario of @xcite , for which there is no magnetic term .
since the proofs in this section are rather lengthy , we have relegated them to appendix [ appb ] and we simply quote the relevant expressions here . the bulk of this section is then devoted to making these results explicit for the case where @xmath452 , @xmath453 and the symplectic form is the magnetic symplectic form derived in the previous section ; see equations and .
the induced poisson structure with interaction terms will turn out to be nothing but the bmr poisson structure ( [ bmrbracket ] ) . secondly
, the poisson map induced by the momentum map @xmath454 is just the shashikanth map described in theorem [ thm : shashi ] , so that the transformed poisson structure is the smbk poisson structure ( [ poisssmbk ] ) . as mentioned above
, the purpose of this section is to generalize the reduction of phase spaces of the form @xmath455 , where @xmath35 is a lie group acting on the symplectic manifold @xmath40 .
this is more general than results in the literature , since the symplectic structure on @xmath455 may contain magnetic terms .
we shall need this generalization in our application .
the proofs of the theorems in this section can be found in appendix [ appb ] .
specifically , if @xmath456 denotes the symplectic form on @xmath40 , we assume that the symplectic structure @xmath201 on @xmath451 is given by @xmath457 where @xmath458 is the canonical symplectic structure on @xmath459 and @xmath460 is a closed two - form on @xmath461 such that @xmath462 , _
i.e. _ the restriction of @xmath460 to @xmath40 is just @xmath463 .
the latter ensures that @xmath201 is a well - defined symplectic form on @xmath451 .
we assume furthermore that @xmath35 acts on @xmath40 and we denote the action of @xmath464 on @xmath465 by @xmath466 .
furthermore , we assume that this action leaves invariant @xmath467 .
the group @xmath35 then acts diagonally on @xmath451 : if @xmath464 and @xmath468 , then @xmath469 [ [ the - momentum - map . ] ] the momentum map .
+ + + + + + + + + + + + + + + + + consider an element @xmath80 of @xmath470 .
we denote the fundamental vector field associated to @xmath80 on @xmath451 by @xmath471 and similar for the fundamental vector field @xmath472 on @xmath473 . under certain topological assumptions ( which are satisfied for the solid - fluid system )
, there exists a momentum map @xmath474 for the action of @xmath35 , defined by @xmath475 where @xmath476 .
it follows from the explicit form for the symplectic form ( [ gensympform ] ) that @xmath454 can be written as @xmath477 , where @xmath478 is the momentum map associated to the canonical symplectic form on @xmath450 : @xmath479 while @xmath480 is the so - called _ @xmath481-potential _ ( borrowing the terminology of @xcite ) : @xmath482 the theory of @xmath481-potentials and poisson reduction for cotangent bundles @xmath450 endowed with a magnetic symplectic form is further developed in @xcite . here
, we are dealing with a product @xmath451 , for which no such theory can be found in the current literature .
[ [ infinitesimal - equivariance - of - the - momentum - map . ] ] infinitesimal equivariance of the momentum map .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + whereas the canonical momentum map @xmath483 is equivariant , the same does not necessarily hold for the @xmath481-potential . to measure non - equivariance
, we introduce a one - cocycle @xmath484 by defining first a family of functions @xmath485 ( where @xmath486 and @xmath487 ) : @xmath488 for all @xmath489 , and then putting @xmath490 .
this definition follows the usual introduction of cocycles for momentum maps ; see for instance @xcite .
the one - cocycle @xmath491 induces a two - cocycle @xmath492 given by @xmath493 where @xmath494 is defined by @xmath495 .
it is not hard to verify that @xmath496 is explicitly given by @xmath497\right > + b(\xi_{h\times p } , \eta_{h\times p}).\ ] ] [ [ the - poisson - structure - on - mathfrakh - ast - times - p . ] ] the poisson structure on @xmath498 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the canonical poisson structure on @xmath499 associated to the symplectic structure ( [ gensympform ] ) gives rise to a poisson structure on the quotient space @xmath500 , which we denote by @xmath501 .
the explicit form of this poisson structure is derived in appendix [ appb ] . before quoting this result ,
we first introduce an operation @xmath502 defined by @xmath503 where @xmath504 is the poisson structure associated to @xmath467 , and an operation @xmath505 by putting @xmath506 using these two operations , the reduced poisson structure is given in the following theorem . note the interaction term due to curvature and the last two terms which are due to the coupling of the lie group @xmath35 with the symplectic manifold @xmath40 .
[ thm : bmr ] the reduced poisson structure on @xmath498 is given by @xmath507 for functions @xmath508 and @xmath509 on @xmath510 .
[ [ shifting - away - the - interaction - terms . ] ] shifting away the interaction terms .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the reduced poisson structure @xmath501 described in theorem [ thm : bmr ] is not canonical , _ i.e. _ it is not the sum of the lie - poisson structure on @xmath511 and the canonical poisson structure on @xmath239 , but rather contains a number of * _ interaction terms_*. however , using the @xmath512-potential @xmath140 , we can define a * _ shift map _ * @xmath513 from @xmath514 to itself , taking @xmath501 into the canonical poisson structure . the price we have to pay for getting rid of interaction terms is the introduction of the non - equivariance cocycle @xmath496 of @xmath140 , as in the following definition .
[ def : natpoisson ] the natural poisson structure on @xmath498 is given by @xmath515 for functions @xmath508 and @xmath509 on @xmath516 .
the main result of this section is then given in the following theorem .
it related the poisson structure with interaction terms with the natural one ( possibly with a cocycle ) .
the former will turn out the bmr poisson structure , while the latter is nothing but the smbk structure .
the cocycle will encode the effects of nonzero circulation on the rigid body . [
thm : shift ] the map @xmath517 given by @xmath518 is a poisson isomorphism taking the poisson structure @xmath501 with interaction terms into the poisson structure @xmath519 : @xmath520 [ [ the - symplectic - leaves . ] ] the symplectic leaves .
+ + + + + + + + + + + + + + + + + + + + + + the shifted poisson structure @xmath519 is the sum of the lie - poisson structure on @xmath521 , the canonical poisson structure on @xmath40 , and a cocycle term . if @xmath522 , this allows us to write down a convenient expression for the symplectic leaves in @xmath523 : [ prop : sympleaf ] for @xmath522 , the symplectic leaves in @xmath498 of the poisson structure @xmath524 are of the form @xmath525 , where @xmath526 is the co - adjoint orbit of an element @xmath527 .
the proof follows that of proposition 10.3.3 in @xcite and relies on the fact that the symplectic leaves are precisely the symplectic reduced spaces .
but since the poisson structure on the reduced space is simply the sum of the lie - poisson and the canonical poisson structure , the reduced space at @xmath150 is @xmath525 , as above . when @xmath528 , we expect the co - adjoint orbit @xmath526 to be replaced by an orbit @xmath529 of a suitable * _ affine _ * action of @xmath35 on @xmath521
now we are ready to specialize the theory in the previous sections to the case of a solid interacting dynamically with point vortices .
recall that our goal is to find the hamiltonian structure for this problem using reduction techniques and to use the shifting map developed in the preceding section to relate the two hamiltonian structures in the literature .
recall that the reduced phase space for the solid - fluid system is @xmath38 .
now , the group @xmath37 acts on @xmath340 by the diagonal left action , denoted by @xmath156 and given in inertial coordinates by @xmath530 hence , @xmath37 acts from the left on @xmath38 using the cotangent lift in the first factor , and thus , this action leaves the hamiltonian ( [ totalhamiltonian ] ) invariant .
this system is of the form @xmath451 , as discussed in the previous section , where @xmath452 and @xmath453 .
we now apply the results of that section to divide out the @xmath37-symmetry and obtain a system on @xmath36 .
this reduction is similar to the passage from inertial to body coordinates for the rigid body ( see for example @xcite ) . to see this , notice that the space @xmath36 is obtained from @xmath38 by dividing out by the diagonal @xmath37-action : the quotient mapping is given by @xmath531 defined as @xmath532 where , if @xmath533 , then @xmath25 and @xmath534 are related by @xmath535 in other words , if @xmath25 describes the location of the @xmath211th vortex in inertial coordinates , then @xmath534 is its location in a frame fixed to the body .
the magnetic symplectic structure is invariant under the action of @xmath37 on @xmath38 described above .
recall the expression ( [ magsympexpl ] ) for @xmath443 , where @xmath328 is given by ( [ betaexpl ] ) .
it is a standard result ( see for instance @xcite ) that the canonical symplectic form @xmath353 on @xmath536 is invariant under the left action of @xmath37 , and a similar result holds for the form @xmath537 . the only thing that remains to be shown is that @xmath436 is @xmath37-invariant , but this follows from the @xmath37-invariance of @xmath538 , which is itself a consequence of proposition [ prop : invariance ] and the fact that @xmath156 and @xmath538 are harmonic conjugates . [ [ the - momentum - map.-1 ] ] the momentum map .
+ + + + + + + + + + + + + + + + + recall from section [ sec : gen ] that the momentum map @xmath454 for the action of @xmath37 on @xmath38 is the difference of two separate parts : @xmath539 , where @xmath483 is the momentum map ( [ mommap ] ) defined by the canonical symplectic form on @xmath536 , while @xmath140 is the so - called @xmath512-potential ( [ bgpot ] ) .
the momentum map @xmath454 is a map from @xmath38 to @xmath233 .
it will be convenient to identify @xmath536 with @xmath540 through left translations , and to use the fact that @xmath233 is isomorphic to @xmath15 , so that @xmath454 is a collection of three functions @xmath541 on @xmath542 .
a typical element of that space will be denoted as @xmath543 , where @xmath533 and @xmath544 , but for the sake of clarity we will usually suppress the argument of @xmath454 . the momentum map @xmath454 associated to the @xmath37-symmmetry represents the spatial momentum of the solid - fluid system , and is given by @xmath545 , where @xmath546 where @xmath547 is the total vortex strength .
the canonical part @xmath483 can be obtained through a standard calculation for cotangent lifted actions ( see @xcite ) .
the result is @xmath548 for @xmath533 and @xmath549 . for the @xmath512-potential , note that the form @xmath328 is the sum of a `` pure vortex '' part ( _ i.e. _ not involving the fluid ) and a part involving the stream functions ( _ i.e. _ the form @xmath436 ) .
the @xmath512-potential corresponding to the pure - vortex part is given by @xmath550 these expressions coincide up to sign with those in @xcite , which is a consequence of the fact that our symplectic structure is the negative of theirs . finally , for the stream function term @xmath551 in @xmath552 , observe that @xmath436 is an @xmath37-invariant one - form on @xmath553 .
the @xmath512-potential associated to @xmath551 is hence given by ( see the remark following theorem 7.1.1 in @xcite ) @xmath554 explicitly , @xmath555 where the elementary stream functions are given by ( [ stream ] ) .
the full momentum map @xmath454 is then the sum of these three contributions : @xmath556 and this is precisely ( [ fullmommmap ] ) .
the expression for the momentum map can be rewritten by introducing the momentum map of the solid - fluid system in * _ body coordinates _ * ( @xcite ) : @xmath557 where @xmath558 are the expression for the elementary stream functions in body coordinates . by using the relation ( [ transition ] ) between inertial and body coordinates , one can show that the spatial and body momentum are related by @xmath559 where @xmath560 .
note that the spatial momentum is a function on @xmath340 , whereas the body momentum is a function on @xmath239 , and that the @xmath37-dependence of @xmath454 is determined by the relation above .
in particular , if we evaluate @xmath454 at the identity , putting @xmath561 and @xmath562 in ( [ relation ] ) , then we obtain just the body momentum : @xmath563 [ [ non - equivariance - of - the - momentum - map . ] ] non - equivariance of the momentum map .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + using the definition ( [ twococycle ] ) , the non - equivariance two - cocycle of the momentum map is a map @xmath564 . to compute @xmath496 , we recall the basis @xmath401 of @xmath64 given in ( [ sebasis ] ) and let @xmath565 be the corresponding dual basis of @xmath233 .
the non - equivariance two - cocycle @xmath496 of the momentum map @xmath454 is given by @xmath566 where @xmath567 is the total vortex strength . in order to use ( [ twococycle ] ) to compute @xmath496 ,
we need the infinitesimal generators corresponding to the basis elements @xmath568 , @xmath569 and @xmath404 . the infinitesimal generator of any element @xmath570 , evaluated at @xmath571 , will be denoted by @xmath572 and is given by @xmath573 where @xmath574 is the infinitesimal generator of @xmath80 for the fundamental action of @xmath37 on @xmath97 .
explicitly , @xmath575 evaluating @xmath328 on these vectors , we obtain @xmath576 remarkably , when calculating @xmath496 , the last two expressions are cancelled entirely by opposite contributions from the remaining terms : @xmath577\right > + \beta_\mu(\tilde{\mathbf{e}}_x , \tilde{\mathbf{e}}_\omega ) \\ & = \phi_y + \beta_\mu(\tilde{\mathbf{e}}_x , \tilde{\mathbf{e}}_\omega ) \\ & = 0 , \end{aligned}\ ] ] and similar for @xmath578 .
the only non - zero term is given by @xmath579 . in its current form , @xmath496 has the same form as the non - equivariance two - cocycle for the @xmath24-vortex problem in an unbounded fluid ( see @xcite ) .
now we come to the conclusion of this paper . using the theory developed in the previous sections ,
we derive an explicit form for the reduced poisson bracket on @xmath36 associated to the magnetic symplectic form ( [ magsympexpl ] ) : this will turn out to be the bmr bracket . in the terminology of section [ sec : gen ] , this is the bracket @xmath501 with interaction terms .
secondly , we then use the shift map @xmath513 ( see ( [ hatshift ] ) ) associated to the momentum map ( [ fullmommmap ] ) to obtain the poisson structure @xmath519 where the interaction terms are absent , at the expense of a non - equivariance cocycle .
when made more explicit , the latter bracket turns out to be the smbk bracket .
[ [ the - bmr - poisson - structure . ] ] the bmr poisson structure .
+ + + + + + + + + + + + + + + + + + + + + + + + + + the bracket obtained by poisson reduction of the magnetic symplectic structure on @xmath580 was described in theorem [ thm : bmr ] .
the explicit computation of this poisson bracket boils down to substituting the explicit expresssion ( [ fullmommmap ] ) for the momentum map into the poisson bracket @xmath581 in theorem [ thm : bmr ] . after a long , but straightforward calculation ,
one obtains the bmr bracket ( [ bmrbracket ] ) . as an illustration , we compute one bracket element , leaving the others to the reader .
we consider the functions @xmath582 and @xmath583 on @xmath220 and compute @xmath584 .
note that , considered as elements of @xmath64 , @xmath585 computing ( [ bmrbracket ] ) term by term , we have first of all that @xmath586 where @xmath236 is the vortex bracket defined in ( [ vortexbracket ] ) . secondly , @xmath587 while the other terms are zero .
finally , it follows that @xmath588 which is precisely the first bracket element of ( [ bmrbracket ] ) .
the computation of the other elements is similar .
[ [ the - smbk - poisson - structure . ] ] the smbk poisson structure .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + by subjecting the bmr poisson structure to the shift map @xmath589 , we can eliminate the interaction terms from the poisson structure , at the expense of introducing a non - equivariance cocycle .
as shown in theorem [ thm : poissonsimple ] , the result is a poisson structure consisting of the sum of the lie - poisson and the vortex poisson structure on the individual factors , together with a cocycle term : @xmath590 the last term is explicitly given by @xmath591 in the case where the total vorticity @xmath23 is zero , this term vanishes and the bracket reduces to the smbk bracket ( [ thm : poissonsimple ] ) . [ [ the - symplectic - leaves.-1 ] ] the symplectic leaves .
+ + + + + + + + + + + + + + + + + + + + + + in the case where the two - cocycle @xmath496 vanishes , or equivalently , when the total circulation @xmath23 is zero , a convenient expression for the symplectic leaves in @xmath36 can be read off from proposition [ prop : sympleaf ] .
recall that the symplectic leaves for the lie - poisson structure in @xmath592 come in two varieties .
one class consists of cylinders whose axis is the @xmath201-axis : @xmath593 while the other class consists of the individual points @xmath594 of the @xmath201-axis .
the symplectic leaves in @xmath36 are then the product of the symplectic leaves of the lie - poisson structure with @xmath239 .
in this paper , we have used reduction theory to give a systematic derivation of the equations of motion for a circular rigid body in a perfect fluid with point vortices . among other things ,
we have derived the poisson structures that govern this problem , and related them to the poisson structures in the literature .
however , the usefulness of our geometric method is not limited to merely reproducing the correct poisson structure for one specific problem : by uncovering fundamental geometric structures such as the neumann connection , we have shed new light on the precise nature of the solid - fluid interaction . moreover , even though a number of simplifying assumptions were made in the beginning , it is expected that the present method can be extended without too much trouble to cover more general cases .
below , we have listed a number of open questions for which our method seems especially suited .
[ [ non - zero - circulation . ] ] non - zero circulation .
+ + + + + + + + + + + + + + + + + + + + + as we have seen in this paper , if the total strength of the point vortices is non - zero , a cocycle term appears in the equations of motion . in a previous paper ( see @xcite ) , we studied the case of a rigid body moving in a potential flow with circulation and found that the circulation manifests itself through the curvature of the neumann connection . from a physical point of view , both phenomena are very similar . indeed , having a non - zero circulation around the rigid body amounts to placing a vortex at the center of the rigid body .
hence , mathematically speaking , there appears to be a duality between the description in terms of cocycles and in terms of curvatures .
having a better mathematical understanding of this duality would prove to be useful for other physical theories as well , for example the dynamics of spin glasses in @xcite .
in addition , it would be interesting to investigate the link between this theory with the work of @xcite .
[ [ arbitrary - body - shapes . ] ] arbitrary body shapes .
+ + + + + + + + + + + + + + + + + + + + + + the pioneering work of smbk was extended to cover the case of rigid bodies with arbitrary shapes by @xcite , and studied further by @xcite with a view towards stability and bio - locomotion . by using methods of complex analysis ,
the methods of the present paper can easily be extended to cover the case of rigid bodies of arbitrary shape as well : any closed curve enclosing a simply - connected area of the plane can be mapped to the unit circle by a suitable conformal transformation and by pulling back the geometric objects in this paper along that map , arbitrary body shapes can be treated . moreover , the objects defined in this paper transform naturally under conformal transformations .
the magnetic form @xmath328 , for instance , will be mapped into a form @xmath595 which has the same appearance as before , but with now @xmath596 the elementary stream functions for a body of that shape . as known from the classical fluid dynamics literature ,
these stream functions also transform naturally under conformal mappings .
[ [ three - dimensional - bodies . ] ] three - dimensional bodies .
+ + + + + + + + + + + + + + + + + + + + + + + + + in contrast to calculations using only standard , ad - hoc methods , our setup naturally generalizes to the case of three - dimensional flows interacting with rigid bodies . @xcite consider a rigid body interacting with * _ vortex rings_*. the space of point vortices @xmath239 then is replaced by the space @xmath597 of @xmath24 vortex rings , and much of the analysis in the 2d case carries through , at the expense of significant computational work . from a geometric point of view
, however , @xmath597 plays a similar role as @xmath239 : both are coadjoint orbits of the group @xmath53 of volume - preserving diffeomorphisms .
hence , @xmath597 is equipped with a natural symplectic structure ( see @xcite ) , which is of importance for the calculation of @xmath328 in this case .
also , the definition of the neumann connection remains valid in three dimensions .
we therefore expect that the geometric approach can be extended without too much difficulty to this case , and will lead to a conceptually much clearer picture .
[ [ routhian - reduction - by - stages - and - dirac - reduction . ] ] routhian reduction by stages and dirac reduction .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + it has been known for some time that the lagrangian analogue of symplectic reduction is * _ routhian reduction _ * ( @xcite ) . by starting from the lagrangian of an incompressible fluid together with the rigid body lagrangian
, one could perform reduction on the lagrangian side using this theory .
the advantage would be that routhian reduction preserves the variational nature of the system .
however , as the reduction procedure for the fluid - solid system consists of two succesive reductions , one would need a theory of * _ routhian reduction by stages_*. developing such a theory would be of considerable interest .
a related approach consists of using dirac structures . through the hamilton - pontryagin principle ,
dirac structures offer a way of incorporating both the lagrangian and hamiltonian formalisms making them ideally suited for systems with degenerate lagrangians ; see @xcite .
since the point vortex lagrangian is degenerate , dirac structures are likely to be useful here .
even though the neumann connection is used in many problems from geometric fluid dynamics or differential geometry , a systematic presentation of its properties seems to be lacking . in this appendix
we prove some basic theorems related to the neumann connection .
recall that if @xmath598 is a principal fibre bundle with structure group @xmath21 , then a connection one - form on @xmath41 is a @xmath294-valued one - form @xmath297 on @xmath41 satisfying the following two properties : 1 .
@xmath297 is * _ @xmath21-equivariant _ * : @xmath599 , where @xmath600 denotes the @xmath21-action , and @xmath601 is the adjoint action on @xmath294 .
2 . let @xmath602 be the * _ infinitesimal generator _ * associated to an element @xmath312 , _ i.e. _ @xmath603 then @xmath604 . at each point
@xmath605 , the connection @xmath297 induces complementary projection operators @xmath606 , given by @xmath607_q(q ) , \quad \text{and } \quad p_h = 1 - p_v,\ ] ] and referred to as the * _ vertical _ * and * _ horizontal _ * projection operators , respectively . more information on principal fiber bundles and connections can be found in @xcite .
[ [ the - curvature . ] ] the curvature .
+ + + + + + + + + + + + + + the * _ curvature _ * of a principal fiber bundle connection @xmath297 is the @xmath294-valued two - form @xmath261 defined as follows .
let @xmath608 and consider vector fields @xmath609 and @xmath610 extending @xmath611 and @xmath612 , _ i.e. _ such that @xmath613 and @xmath614 .
denote the horizontal part of @xmath609 as @xmath615 , and similar for @xmath616 .
then , @xmath261 can be conveniently expressed as @xmath617).\ ] ] with this definition , @xmath261 is @xmath21-equivariant and vanishes on vertical vector fields : @xmath618 for all @xmath619 and @xmath312 .
[ [ the - connection - one - form.-1 ] ] the connection one - form .
+ + + + + + + + + + + + + + + + + + + + + + + + we recall from section [ sec : neumann ] that the connection one - form for the neumann connection is given by @xmath620 where @xmath158 is the divergence - free part in the helmholtz - hodge decomposition ( [ hhdecomp ] ) of the eulerian velocity @xmath117 ; see ( [ connection ] ) . before showing that this prescription indeed yields a well - defined connection form
, we note that each divergence - free vector field @xmath183 defines a vertical vector field on @xmath41 , denoted by @xmath621 and given by @xmath622 the vector field @xmath621 is the infinitesimal generator associated to @xmath117 under the action of @xmath53 on @xmath41 .
[ prop : conn ] the one - form @xmath297 defined in ( [ connection ] ) is a connection one - form . in other words , 1 .
@xmath297 is @xmath53-equivariant : for all @xmath623 and @xmath624 , @xmath625 2 . if @xmath117 is an element of @xmath179 with associated fundamental vector field @xmath621 , then @xmath626 . to prove equivariance , note that the eulerian velocity associated to @xmath139 is equal to that associated to @xmath138 : @xmath627 hence
, @xmath628 to prove the second property , note that the push - forward @xmath629 is divergence - free and tangent to the boundary of @xmath30 since @xmath117 is divergence - free and tangent to @xmath630 and @xmath101 is volume - preserving . evaluating the connection one - form on @xmath117 therefore gives @xmath631 which concludes the proof that @xmath297 is a well - defined connection one - form .
[ [ the - horizontal - lift . ] ] the horizontal lift .
+ + + + + + + + + + + + + + + + + + + + the horizontal lift operator associates to each @xmath632 a linear map @xmath633 with the following properties : 1 . the composition @xmath634 is the identity in @xmath281 ; 2 .
the image of @xmath635 consists of horizontal vectors : @xmath636 . for the neumann connection ,
the horizontal lift is given by @xmath637 where @xmath156 is the solution of the neumann problem ( [ neumann ] ) associated to @xmath253 ; _ i.e. _ @xmath638 .
it is easy to show that @xmath639 satisfies both properties of the horizontal lift , and hence that @xmath639 is indeed the horizontal lift of the neumann connection as we claimed in section [ sec : neumann ] .
[ [ the - neumann - connection - as - a - mechanical - connection . ] ] the neumann connection as a mechanical connection .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + when given a group - invariant metric on the total space of a principal fiber bundle , the * _ mechanical connection _ * is defined by declaring its horizontal subspaces to be orthogonal to the vertical bundle .
the terminology stems from the fact that the underlying metric is usually the kinetic - energy metric of a mechanical system . since there exists a natural metric ( [ metric ] ) on @xmath41 , it should therefore come as no surprise that the neumann connection can also be viewed as a mechanical connection .
the horizontal subspace of the neumann connection at an element @xmath640 is given by the kernel of @xmath641 .
explicitly , a tangent vector @xmath642 is in the kernel of @xmath641 if and only if @xmath643 , where @xmath156 is the solution of the neumann problem ( [ neumann ] ) associated to @xmath253 .
on the other hand , the vertical subspace at @xmath632 is generated by elements of the form @xmath644 .
verifying that the neumann connection is indeed the mechanical connection now boils down to checking that the vertical and horizontal subspaces are orthogonal with respect to the metric ( [ metric ] ) .
let @xmath645 and @xmath644 be horizontal and vertical tangent vectors , respectively .
then we have @xmath646 where the last equality follows from the @xmath278-orthogonality between gradient vector fields and divergence - free vector fields tangent to the boundary of @xmath30 .
in this technical appendix , we provide proofs for the theorems in section [ sec : gen ] regarding the reduced poisson structures with or without interaction terms .
we do so by using the diffeomorphism of @xmath499 with @xmath647 given by left translation .
more specifically , we introduce a diffeomorphism @xmath648 by @xmath649 and a diffeomorphism @xmath650 by @xmath651 such that the following diagram commutes : @xmath652^\lambda \ar[d]_{\pi_h } & h \times \mathfrak{h } ^{\ast } \times p \ar[d]^{\mathrm{pr } } \\ h \times p \ar[r]_\lambda & h \times p}\ ] ] here , @xmath653 is given by @xmath654 , while @xmath655 simply forgets the second factor : @xmath656 .
the advantage of this construction is that under @xmath657 the action ( [ diagaction ] ) becomes left translation on the first factor .
that is , if @xmath658 , then @xmath659 .
the fact that the diagonal action ( [ diagaction ] ) reduces in this way greatly simplifies poisson reduction .
we start by deriving in lemma [ lemma : sympbody ] an explicit expression for the push - forward @xmath660 of the symplectic form .
we then define a natural * _ shift map _ * @xmath153 from @xmath661 to itself , given by @xmath662 where @xmath480 is the @xmath512-potential associated to the action of @xmath35 on @xmath663 , _
i.e. _ the solution of ( [ bgpot ] ) .
this map takes the symplectic structure ( [ bigsympstruct ] ) with interaction terms of lemma [ lemma : sympbody ] into one where the interaction terms are absent .
the explicit form of the resulting symplectic form is derived in lemma [ lemma : pf ] .
[ lemma : sympbody ] the push - forward @xmath660 of the symplectic form @xmath201 to @xmath668 is given by @xmath669\right > \label{bigsympstruct }
\\ - b(e , x)(\xi_{h\times p}(e , x ) , \eta_{h\times p}(e , x ) ) - b(e , x)((0 , v_x ) , ( 0 , w_x ) ) \\ - \left<\mathbf{d } \phi_\xi(e , x ) , ( 0 , w_x)\right > + \left<\mathbf{d } \phi_\eta(e , x ) , ( 0 , v_x)\right > , \end{gathered}\ ] ] where @xmath670 is an element of @xmath668 and @xmath671 , @xmath672 are elements of @xmath673 .
notice first that @xmath674 . the first term is the expression for the canonical symplectic form in body coordinates calculated by r. cushman ( quoted in @xcite , proposition 4.4.1 ) and is given by @xmath675\right>.\ ] ] for the second term , we have @xmath676 where @xmath677 denotes the action of @xmath35 on @xmath40 : @xmath678 , and @xmath679 is the map defined by @xmath680 .
note that @xmath681 . taking this into account , as well as the @xmath35-invariance of @xmath460 ,
allows us to rewrite the above expression as @xmath682 the third term can be rewritten as @xmath683 and similar for the last term . putting everything together ,
we obtain ( [ bigsympstruct ] ) .
[ lemma : pf ] the push - forward @xmath684 on @xmath685 is again a symplectic form , given by @xmath686 here , the arguments of @xmath684 have the same meaning as in lemma [ lemma : sympbody ] , and @xmath496 is the non - equivariance two - cocycle ( [ twococycle ] ) of @xmath140 .
the pushforward of @xmath687 under @xmath153 is given by @xmath688\right>,\end{gathered}\ ] ] where @xmath689 substituting this into ( [ pushforwardcushman ] ) , we get @xmath690 and similar for @xmath691 . for the term involving @xmath692 , we obtain @xmath693\right > & = \left<\mu , [ \xi , \eta]\right > + \left<\phi(e , x ) , [ \xi , \eta]\right > \\ & = \left<\mu , [ \xi , \eta]\right > - \sigma(\xi , \eta ) + b(e , x)(\xi_{h \times p}(e , x ) , \eta_{h \times p}(e , x ) ) , \end{aligned}\ ] ] where we have used the definition of @xmath496 .
hence , @xmath694 the last three terms are just the interaction terms . hence , by substituting this expression into the expression for @xmath695 , these terms cancel out , leaving ( [ pfomega ] ) .
let @xmath670 be an element of @xmath668 and write @xmath705 as @xmath706 , where @xmath707 , @xmath708 , and @xmath709 .
note also that @xmath710 is of the form @xmath711 where @xmath712 denotes the differential of @xmath713 keeping @xmath62 and @xmath150 fixed , and similar for @xmath714 .
a long but straightforward calculation shows that @xmath715 where the map @xmath716 is given by @xmath717 iff @xmath718 , while @xmath719 and @xmath720 and @xmath80 are given by ( [ components ] ) .
here , @xmath721 is the map defined by @xmath722 , for all @xmath723 and @xmath724 .
note that @xmath725 the poisson bracket is then given by @xmath726 the first term is equal to @xmath727 while for the second term , we have @xmath728 using the definition of the induced poisson structure in terms of the original symplectic structure : @xmath729 .
writing out the last term requires a little more work .
denote @xmath730 the last term then becomes @xmath731 the first term on the right - hand side can be rewritten as @xmath732 putting everything together , we obtain ( [ eq : poisson ] ) .
having established the explicit form for the interaction poisson structure @xmath733 on @xmath661 , we can at last show that the reduced poisson structure on @xmath498 is given by expression ( [ redpoisson ] ) in theorem [ thm : bmr ] .
according to the poisson reduction theorem ( see @xcite ) , the reduced poisson structure @xmath501 on @xmath498 is defined by the prescription @xmath734 where @xmath733 is the poisson structure on @xmath696 described in proposition [ prop : poisson ] and @xmath735 forgets the first factor : @xmath736 . in other words ,
the reduced poisson structure is the original poisson structure ( [ eq : poisson ] ) evaluated on functions @xmath508 and @xmath509 which do not depend on @xmath487 .
this yields the expression ( [ redpoisson ] ) for @xmath501 .
the symplectic form @xmath737 and the poisson structure @xmath501 are characterized by the presence of * _ interaction terms_*. by this , we mean that the symplectic structure on @xmath647 is not simply the sum of the symplectic form @xmath687 on @xmath738 and the symplectic form @xmath467 on @xmath40 , but contains terms involving the curvature @xmath460 and the @xmath512-potentials @xmath739 , and similar for @xmath501 .
as shown in lemma [ lemma : pf ] , one can remove the interaction terms from @xmath737 at the expense of introducing a cocycle @xmath496 by using the shift map @xmath513 .
the resulting symplectic structure then induces a reduced poisson structure on @xmath500 , which is just the natural poisson structure of definition [ def : natpoisson ] .
[ thm : poissonsimple ] the symplectic form @xmath684 on @xmath668 is @xmath35-invariant .
the poisson structure on @xmath498 induced by @xmath684 is the natural poisson structure @xmath519 of definition [ def : natpoisson ] .
the map @xmath54 shifts away the interacting terms in the symplectic form , and induces a map on the lie algebra level .
it is easy to see that this map is just the map @xmath513 defined in ( [ hatshift ] ) .
using this observation , the proof of theorem [ thm : shift ] now follows immediately .
the map @xmath589 is the quotient space map induced by @xmath153 , which is a symplectic map taking the symplectic form ( [ bigsympstruct ] ) with interaction terms into the symplectic form ( [ pfomega ] ) without interaction terms .
hence , the induced map is a poisson map with respect to the induced poisson structures , _ i.e. _ ( [ poissonmap ] ) holds .
abraham , r. and j. e. marsden [ 1978 ] , _ foundations of mechanics_. benjamin / cummings publishing co. inc .
advanced book program , reading , mass .
second edition , revised and enlarged , with the assistance of tudor raiu and richard cushman .
adams , m. and t. ratiu [ 1988 ] , the three - point vortex problem : commutative and noncommutative integrability . in _
hamiltonian dynamical systems ( boulder , co , 1987 ) _ , volume 81 of _ contemp .
_ , pages 245257 .
soc . , providence , ri .
cendra , h. , j. marsden , and t. s. ratiu [ 2004 ] , cocycles , compatibility , and poisson brackets for complex fluids . in _ advances in multifield theories for continua with substructure _ , model .
technol . , pages 5173 .
birkhuser boston , boston , ma .
shashikanth , b. n. [ 2005 ] , poisson brackets for the dynamically interacting system of a 2d rigid cylinder and @xmath24 point vortices : the case of arbitrary smooth cylinder shapes , _ regul .
chaotic dyn .
_ * 10 * , 114 .
shashikanth , b. n. , j. e. marsden , j. w. burdick , and s. d. kelly [ 2002 ] , the hamiltonian structure of a two - dimensional rigid circular cylinder interacting dynamically with @xmath24 point vortices , _ phys .
fluids _ * 14 * , 12141227 .
shashikanth , b. n. , a. sheshmani , s. d. kelly , and j. e. marsden [ 2008 ] , hamiltonian structure for a neutrally buoyant rigid body interacting with n vortex rings of arbitrary shape : the case of arbitrary smooth body shape , _ theor .
fluid dyn . _
* 22 * , 3764 .
vankerschaver , j. , e. kanso , and j. e. marsden [ 2008 ] , _ the motion of solid bodies in potential flow with circulation : a geometric outlook _ , to appear in the proceedings of the asme 2008 dynamic systems and control conference , 2008 . | we derive the equations of motion for a planar rigid body of circular shape moving in a 2d perfect fluid with point vortices using symplectic reduction by stages . after formulating the theory as a mechanical system on a configuration space which is the product of a space of embeddings and the special euclidian group in two dimensions , we divide out by the particle relabelling symmetry and then by the residual rotational and translational symmetry .
the result of the first stage reduction is that the system is described by a non - standard magnetic symplectic form encoding the effects of the fluid , while at the second stage , a careful analysis of the momentum map shows the existence of two equivalent poisson structures for this problem . for the solid - fluid system
, we hence recover the ad hoc poisson structures calculated by shashikanth , marsden , burdick and kelly on the one hand , and borisov , mamaev , and ramodanov on the other hand . as a side result , we obtain a convenient expression for the symplectic leaves of the reduced system and we shed further light on the interplay between curvatures and cocycles in the description of the dynamics . |
the study of flares on m dwarfs has a rich history , spanning more than seven decades ( e.g. * ? ? ?
* ; * ? ? ?
the similarity of these events to resolved flares on the sun was recognized many years ago ( e.g. * ? ? ?
the seminal paper by @xcite laid the groundwork for our present understanding of the broad multi - wavelength properties of m dwarf flares .
the basic two - component flare spectral model ( h recombination + warm blackbody continuum ) presented in @xcite is useful for estimating the total energy and stellar surface extent of flares in the optical regime @xcite .
however , this model does not specifically address the detailed spectral features observed during flares , such as line emission from h , he i , he ii , and ca ii h&k , nor the evolution of the spectral energy distribution ( sed ) as the flare cools and decays .
a partnership of multi - wavelength photometric and spectroscopic time - domain observing campaigns ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and detailed radiative hydrodynamical modeling @xcite are required to extend our understanding of stellar flare emission physics .
@xcite found that the fraction of m dwarfs that show evidence of magnetic activity as defined by h@xmath9 emission decreases with height above the galactic plane .
this is thought to be an age effect , with the older , less active stars lying further from the plane due to dynamical heating and radial migration within the disk @xcite .
flares may play an important role in the main sequence stellar evolution for low - mass stars , and correlating m dwarf flare rates with mass and metallicity promises to yield an effective constraint on the ages of stellar populations .
models and observations of the initial mass function indicate that m dwarfs are the primary stellar component in the galaxy by number @xcite .
recent large scale surveys , such as the sloan digital sky survey ( * ? ? ?
* hereafter sdss ) , have photometrically identified millions of m dwarfs @xcite .
these stars make excellent tracers of nearby galactic structure due to their high number densities @xcite , and future survey missions , like the large synoptic survey telescope ( lsst ; * ? ? ?
* ) will detect m dwarfs to nearly the full extent of the galaxy , using longer wavelength bandpasses and deep repeat imaging .
lsst will map the sky approximately 1000 times over 10 years using @xmath10 bands , and is expected to detect flares both as variability in previously quiescent stars , and as transient emission from previously undetected sources .
this foreground `` fog '' of characteristically blue transient events will prove a significant contaminant to any statistical measurements of variability from cosmological sources @xcite .
* hereafter k09 ) used the sdss stripe 82 time domain photometric data @xcite to look for flares from @xmath050,000 m dwarf light curves in the @xmath3 and @xmath11-bands . from these sparse light curves they were able to recover a flare rate consistent with those found from dedicated photometric monitoring campaigns ( e.g. * ? ? ?
the rate of flares was found to decrease with peak flare luminosity , and increase with stellar spectral type ( redder colors ) .
@xcite found serendipitous flares in the time - resolved sdss spectroscopic data from @xmath038,000 m dwarfs across the entire sdss data release 6 footprint @xcite .
they confirmed the increase of the flare duty cycle with later spectral type , and further investigated the galactic height dependence of flaring activity , finding that the stars that flared were even more closely confined to the galactic plane than the stars with h@xmath9 activity .
the few studies of m dwarf flares in the near infrared ( nir ) that have been previously attempted have not found a consistent relationship to the optical emission .
@xcite , for example , measured a marginal anti - correlation between optical u - band and infrared k - band photometry of two flares .
similarly , @xcite found no k - band enhancement during 26 m dwarf flares that were observed simultaneously in the u - band .
observations of flares on the sun in the nir offer a more promising avenue .
@xcite studied the luminosity of solar flares at 1.6@xmath12 m , the h@xmath13 opacity minimum , and found a correlation with hard x - ray and white - light optical emission for two x - class flares .
their data indicated that the nir band reached peak luminosity after the x - ray , and was responding to secondary heating , possibly from chromospheric back - warming .
modern m dwarf flare studies are expanding both in wavelength and temporal coverage to gain insight into stellar flare physics , as well as the contaminating impact these events have on detecting other sources of variability .
@xcite found a correlation between flare energies in the optical , near - uv , and x - ray . @xcite
have observed flaring m dwarfs with nir spectroscopy to characterize the correlation between nir emission lines , such as the h paschen series , and blue optical broadband flux during flares .
dedicated high time - cadence white - light flare studies from the kepler mission @xcite will allow an order - of - magnitude improvement in the study of flare frequency distributions @xcite and their possible evolution .
@xcite have undertaken the most comprehensive simultaneous optical and nir photometric observing campaign for flares to date to determine the effects of flares on exoplanet detection around m dwarfs .
these high cadence and high precision data place upper limits on the broadband nir emission from several medium @xmath3-band flares from m3 and m4 stars with @xmath03.9 milli - mag photometric precision . in this paper
we seek to extend the analysis of k09 using the time domain data from the 2mass calibration point source working database ( hereafter cal - pswdb ) as well as the sdss stripe 82 data .
the combination of these data allows us to statistically characterize for the first time the rate of broadband m dwarf flares in red optical and nir wavelengths .
our data are outlined in 2 .
variability statistics are discussed in 3 .
flare rates as a function of both spectral type and observed passband are presented in 4 . a spectroscopic model of a m dwarf flare
is given in 5 , and the predicted sensitivity to finding flares in each band in 6 .
finally a concluding discussion is given in 7 .
our data are drawn from the sdss and 2mass photometric time domain databases .
every m dwarf in our study has single - epoch sdss data release 7 ( dr7 * ? ? ?
* ) photometry , used to determine photometric spectral types and distances . here
we describe the selection criteria and spatial matching of our time domain samples .
our near infrared data come from the 2mass cal - pswdb .
full details are given in the online explanatory supplement and cutri et al .
( 2006 ) . in summary ,
the repeat imaging was obtained to provide consistent photometric calibration for the 2mass survey over the four years of its operation .
there were 35 regular calibration tiles spread across the sky , with each tile spanning approximately 85 in ra , and 1@xmath14 in dec .
one of these 35 tiles was visited hourly during survey operations by the 2mass telescopes , with simultaneous imaging in the @xmath15 bands .
this sequence produced between 562 and 3692 epochs for these calibration fields .
five additional tiles were imaged in the lmc and smc with fewer epochs .
a total of 191 million point source measurements are contained in the database . a pilot study with
the cal - pswdb by @xcite revealed a 2.6 day low - mass eclipsing binary system .
@xcite used the cal - pswdb data to search for periodic variability of low mass stars , finding several eclipsing binary candidates .
additionally , @xcite found a periodic signature in cal - pswdb from a possible triple young stellar object in the @xmath16 ophiuchus star forming region .
we entered the cal - pswdb point source measurements into a mysql database to allow rapid matching of these data to other datasets .
the 191 million measurements were clustered into individual objects using the spatial clustering algorithm , optics @xcite , creating light curves for 113,030 unique objects .
of the 35 primary cal - pswdb tiles 16 were coincident with the sdss data release 7 photometric catalog ( * ? ? ?
* hereafter dr7 ) , and we found 11,445 matched point source objects using a 25 radius .
figure [ radec ] shows the positions of the 40 cal - pswdb tiles as compared to the sdss photometric footprint .
an optical nir @xmath17 color - color diagram for the matched point source objects is shown in figure [ colorcolor ] , with 1,963 galaxies and 9,369 stars .
the sdss obj_type flag was used to distinguish between stars and galaxies .
we removed the 113 objects with unknown object type ( obj_type=0 ) .
the distribution of stars in our sample is well fit by the @xcite fiducial stellar locus .
we imposed magnitude limits of @xmath18 and @xmath19 , and standard sdss photometric flag cuts to ensure good sdss dr7 photometry .
m dwarfs were selected and spectral types assigned using @xmath1 and @xmath2 colors and the covariance matrix technique as outlined in k09 .
we found 4,860 stars with photometric spectral types later than m0 , with very few stars in the m7-m9 range due to their faint apparent magnitudes .
the resulting 4,731 m0m6 stars had a total of 5.75@xmath20 point source measurements in cal - pswdb .
only 12 of our stars were in the sdss spectroscopic m dwarf catalog @xcite so we rely solely on the sdss photometry for classification .
11,445 ( objects ) & & cal - pswdb
dr7 match + 2 .
& 5.75@xmath20 ( epochs ) 4,731 ( stars ) & @xmath21sptype @xmath22 , obj_type = 6 & m dwarfs only + 3 . & 2.51@xmath20 ( epochs ) 1,321 ( stars ) & ph_qual = a , & high s / n epochs + & & @xmath23 & deblending + 4 . &
1.14@xmath20 ( @xmath24 epochs ) & @xmath25 , @xmath26 , @xmath27 , @xmath28 & good photometry + 5 . & 9.08@xmath29 ( @xmath30 epochs ) & @xmath26 , @xmath31 , @xmath32 , @xmath33 & good photometry [ sampleselection ] we restricted our analysis to epochs with the cal - pswdb flag ph_qual = a , which required every @xmath34 measurement to have a signal - to - noise ratio of at least 10 . to ensure that each dr7 source matched to only a single cal - pswdb object , we conservatively removed all objects with a spatial standard deviation in their cal - pswdb data of @xmath35 .
this cut limited the effects of source confusion and de - blending , but also removed any high proper motion objects from our sample .
we additionally excluded epochs where photometric errors exceeded @xmath36 , @xmath37 , or @xmath38 .
our final cal - pswdb sample contained 1,321 m dwarfs with @xmath02@xmath20 epochs , for an average of @xmath01,900 good epochs per star .
the selection of our cal - pswdb sample is detailed in table [ sampleselection ] .
in addition to the matched time domain @xmath34 data from cal - pswdb , we analyzed m dwarfs from the sdss stripe 82 repeat scan database @xcite to obtain time domain data in the @xmath39 bands @xcite .
these data were previously used by k09 to search for flares in the @xmath3- and @xmath11-bands , and we adopt their sample as described in their table 2 .
briefly , these data contain 50,130 photometrically selected m dwarfs , with 1.9 million total photometric measurements .
the stars were required to have good @xmath3-band magnitudes ( @xmath40 ) , which limited the sample to relatively bright m dwarfs .
only 429 of these objects have matches in the cal - pswdb database , due to the limited spatial overlap .
we measure the flare rate independently in each filter to build statistics using the entire population from each time - domain survey . the spatial location of the sdss equatorial stripe 82 data is shown in figure [ radec ] in blue .
figure [ lc ] gives an example of the eight - band light curves we have produced for a representative m0 star with both stripe 82 and cal - pswdb photometry , illustrating the differences in time coverage and cadence between the two surveys .
this star had 2,922 epochs of data in each of the @xmath34 bands , and 70 epochs in each of @xmath39 .
the time units for all of our sdss and 2mass data have been converted to barycentric dynamical time ( @xmath41 ) , and are presented as a modified barycentric julian date ( bjd-2400000.5 ) .
figure [ sptype ] shows the numbers of stars in each spectral type bin for our stripe 82 and cal - pswdb samples .
discerning true variability from background noise requires sophisticated statistical techniques . following k09 ,
we utilize a modified welch stetson variability index , @xmath42 , to search for flares in our sample @xcite .
this statistic uses measurements in two bands to separate real stellar variability from the intrinsic scatter in a light curve . for the @xmath43-th measurement of a light curve in @xmath7- and @xmath44-bands , for example ,
@xmath42 is defined as : @xmath45\left [ \frac{m_h(n ) -
\langle m_h\rangle}{\sigma_h(n)}\right]\,,\ ] ] where @xmath46 is the apparent magnitude at epoch @xmath43 , @xmath47 the median magnitude over all epochs , and @xmath48 the photometric error at epoch @xmath43 .
when the magnitude increases or decreases for _ both _ bands simultaneously , @xmath42 yields a positive value .
conversely , when the magnitude from one band is higher than normal while the other is diminished , @xmath42 is negative . for purely random noise between the two filters , this index would yield a symmetrical distribution about @xmath49 .
excesses of positive @xmath42 values represent correlated variability , from events such as flares or eclipses .
we computed the @xmath42 statistic for all pairs of bandpasses `` adjacent '' in wavelength space ( e.g. @xmath50 , @xmath51 for every epoch in our samples .
since the sdss and 2mass data were not obtained contemporaneously ( see figure [ lc ] ) we do not compute @xmath52 .
figure [ phi ] shows the @xmath42 histograms for all objects in our stripe 82 and cal - pswdb samples .
the black line indicates the @xmath53 data , hereafter denoted @xmath54 , while the blue line shows the @xmath55 data , hereafter @xmath56 . as the sampling cadence of stripe 82 and cal - pswdb surveys was much longer than typical flare timescales of minutes to hours @xcite , our flare search relies on single - epoch outliers in the light curves . in order to act as an effective variability discriminant the two bandpasses making up the @xmath42 statistic must be imaged nearly simultaneously to capture the same physical event .
the 2mass @xmath34 bands were imaged simultaneously @xcite , allowing the photometry to be directly compared for each 7.8 second exposure .
the sdss used a drift - scanning technique that produced separate 52.4 second exposures in each filter taken in the order @xmath57 @xcite . as a result the @xmath58 values , for example , come from measurements which were separated by only @xmath01 minute , while the @xmath59 values are derived from data which were separated by almost 5 minutes .
we searched for flares in each of the six @xmath42 distributions separately , but recognize the limitation that flares with timescales of only a few minutes may not have been imaged by all sdss bands .
as such , these rapid events would not yield large @xmath42 values , and may fall below our detection thresholds . to remove large @xmath42 values from the sample that resulted from a
_ decrease _ in flux in both bands , for example due to photon loss from clouds or eclipses , we required all epochs in our investigation to have @xmath60 for the bluer ( shorter wavelength ) filter in each of the @xmath54 and @xmath56 distributions .
this removed approximately half the epochs in each distribution , as expected .
as discussed in k09 , the @xmath56 distribution , also referred to as the `` null '' distribution , is representative of uncorrelated noise in the light curve .
the @xmath56 distribution was still well modeled by gaussian noise in all filters after the removal of half the epochs .
comparison of the @xmath54 distribution with the @xmath56 yields excesses in the @xmath54 due to candidate flaring epochs .
in this section we explore the selection of flares from our light curves , and the luminosities of the flare candidate epochs .
as discussed above , given our sparse sampling we assume that each flare is observed in only one epoch of our data .
a threshold value for @xmath54 must be chosen from the null distributions to separate flare epochs from the quiescent epochs with statistical fluctuations .
if too high a threshold is chosen , we may lose valuable weak signal in the redder bandpasses by disregarding real flare events .
however , too low a threshold will add significant false positives to the resulting flare rates . using the false discovery rate method @xcite on the @xmath61 distribution
, k09 determined that a conservative cutoff for flares was @xmath62 .
this ensured less than 10% contamination from non - flaring epochs would be included in their final analysis .
we performed the false discovery rate analysis to limit the contamination to 10% in all six of our @xmath56 distributions , and determined @xmath42 thresholds of ( 100 , 100 , 120 , 500 , 50 , 50 ) for @xmath63 respectively .
the @xmath56 distributions using 2mass filters in figure [ phi ] decline rapidly at @xmath64 , nearly an order of magnitude lower values than for the optical filters .
this is indicative of the lower intrinsic stellar variability seen at these wavelengths , and the larger photometric errors in the 2mass survey .
the @xmath42 thresholds are shown as dashed vertical lines in figure [ phi ] .
the number of flaring epochs for each filter combination was very small compared to the total number of observations .
previous studies such as k09 have shown that flare rates for m dwarfs increase as a function of spectral type . to examine this trend we grouped our sample into three spectral type bins ( m0-m1 , m2-m3 , m4-m6 ) .
the ratio of flare candidate epochs to the total number of epochs for each bin is shown in figure [ ffrac ] . as expected ,
all the optical sdss filter combinations show an increasing numbers of flares with later spectral type ( k09 ) .
there were considerably fewer flare epochs detected in the nir passbands , although the @xmath65 and @xmath66 flare rates fall within the range of the sdss data in figure [ ffrac ] .
this is likely due to the weak signatures of flares in these nir bands , as discussed in 5 .
table [ flaretable ] presents the total numbers of flares in each band for the three spectral type bins .
lcccc ug & m0-m1 & 81 & 1.56 & 5.4@xmath67 + ug & m2-m3 & 116 & 6.85 & 1.7@xmath68 + ug & m4-m6 & 262 & 1.15 & 2.3@xmath69 + gr & m0-m1 & 722 & 1.66 & 4.6@xmath68 + gr & m2-m3 & 322 & 7.25 & 4.5@xmath68 + gr & m4-m6 & 251 & 1.25 & 2.1@xmath69 + ri & m0-m1 & 567 & 1.66 & 3.6@xmath68 + ri & m2-m3 & 283 & 7.35 & 3.9@xmath68 + ri & m4-m6 & 149 & 1.25 & 1.2@xmath69 + iz & m0-m1 & 125 & 1.66 & 8.0@xmath67 + iz & m2-m3 & 100 & 7.25 & 1.4@xmath68 + iz & m4-m6 & 53 & 1.25 & 4.3@xmath68 + jh & m0-m1 & 41 & 6.35 & 6.5@xmath67 + jh & m2-m3 & 10 & 3.05 & 3.2@xmath67 + jh & m4-m6 & 48 & 2.05 & 2.4@xmath68 + hk & m0-m1 & 99 & 5.15 & 1.9@xmath68 + hk & m2-m3 & 13 & 2.45 & 5.5@xmath67 + hk & m4-m6 & 69 & 1.75 & 4.1@xmath68 [ flaretable ] threshold divided by the total number of good epochs , for each filter combination and spectral type bin.,width=336 ] to compare our results in many photometric bands to those of previous studies , we must transform the flares from apparent magnitude enhancements to luminosities .
the apparent magnitudes for every epoch and filter in our time - domain stripe 82 and cal - pswdb samples were first converted to observed flux , in units of erg s@xmath70 @xmath71m@xmath70 , using the calibrations from @xcite and @xcite respectively . to estimate the quiescent luminosity for each epoch , we obtained distance estimates for our stars from the @xmath72 photometric parallax relation of @xcite , using the @xmath73 color from dr7 .
errors in the computed luminosities are dominated by the distance uncertainties .
many previous studies have shown that the frequency of flares decreases as a function of increasing flare energy ( e.g. * ? ? ? * ) or flare luminosity ( k09 ) . for each bandpass we selected the epochs identified as flare candidates in table [ sampleselection ] , and found a cumulative distribution in flare luminosity . the frequency per hour that these flares are observed was calculated by dividing the cumulative number distribution by the total observing time in hours for each filter , giving the flare frequency distributions plotted in figure [ ffd ] .
these show for each filter , @xmath74 , the number of flares per hour of observation , @xmath75 , as a function of observed luminosity , @xmath76 . the yz cmi flare frequency distribution from @xcite
is shown , and corresponds well with our optical distributions .
the optical ( @xmath39 ) distributions are very similar , indicating that flares produce smooth continuous variations in these bandpasses . as with figure [ ffrac ] ,
the @xmath34-band flare frequency distributions in figure [ ffd ] are highly under - sampled .
nir flares were recovered more than two orders of magnitude less frequently than for sdss optical colors for the m4m6 bin .
the turnover at low luminosities seen in the sdss filters is due to incompleteness in the low - luminosity flare census . to explore whether the observed power - law distributions ( from log @xmath76=29 to 31 erg s@xmath70 ) extend to even lower luminosity flares than our @xmath42 thresholds recover , we moved the @xmath54 detection thresholds to higher and lower values in the @xmath3-band . in figure [ phicut ]
the power - law slope appears to extend to much lower luminosities as a result of the lower @xmath42 thresholds .
we recovered many additional flare candidate epochs , but with a much higher rate of false positives .
the high - luminosity end of figure [ phicut ] ( above log @xmath7729.1 erg s@xmath70 ) is insensitive to the choice of @xmath42 threshold .
we present our flare frequency distributions as observed , but note that it is possible to correct the frequencies for the efficiency in recovering flares as a function of luminosity ( e. j. hilton 2011 phd thesis ) .
the @xmath42 distributions in figure [ phi ] indicate that there are many more epochs in the @xmath54 distribution than in the @xmath56 , showing that the two bandpasses which make up each @xmath42 distribution preferentially increased in flux together .
the excess between the @xmath54 and @xmath56 is also evident at values of @xmath42 well below the flare candidate thresholds . by subtracting the @xmath54 from the @xmath56 distributions for each of the six filter combinations in figure [ phi ] we are left with residual @xmath54 distributions , as shown in figure [ resid_ri ] for @xmath58 .
these residual distributions follow a roughly power - law profile below the @xmath54 thresholds , with a turn - over at low @xmath54 amplitudes .
we fit the residual distributions using the equation @xmath78 where @xmath79 is the @xmath80 residual , @xmath9 is the peak amplitude for the residual , @xmath81 a power - law slope at larger luminosities , @xmath82 is the turnover @xmath42 between the power - law and the flattening at low @xmath42 , @xmath83 the exponential plateau slope at low @xmath42 , and @xmath84 is the @xmath54 value .
figure [ resid_ri ] also illustrates where in the residual distribution each of the fit coefficients constrains .
this functional form was chosen to simultaneously fit the residual power - law tail at large @xmath42 , and the turnover at low @xmath42 values due to random noise dominating .
the power - law coefficient @xmath81 is plotted for each of the distributions in fig [ resid ] .
the @xmath81 values ranged between -1.9 and -3.1 . the higher slope in @xmath81 in the @xmath59 fit
may be due to the better photometric precision in the @xmath11 and @xmath85 bands , compared to the @xmath3 band , allowing those filters to be more sensitive to flares below the @xmath42 threshold .
the cal - pswdb @xmath86 conversely shows the smallest observed residual , again suggesting that these longer wavelength filters are less affected by intrinsic stellar variability
. however , all the fits are consistent with a value of @xmath87 within the errors . residual ( black ) , and the corresponding fit ( red ) defined in eqn [ phieqn].,width=336 ] the physical cause of this power - law slope , @xmath81 , at @xmath42 values less than our flare cutoff threshold
, may be the extension of the increasing flare frequency at lower luminosities .
@xcite showed that the flare frequency distribution has a power - law shape , with a turnover at lower energies due to confusion with the quiescent variability , and a fall - off at higher energies due to incompleteness .
their flare detection extended to lower energy flares than our analysis was sensitive to .
this trend has been verified for m dwarf flares more recently by k09 and @xcite .
there is not a direct transformation to compare between our observed @xmath42 residual distributions and the flare frequency distributions calculated by @xcite and others .
our residual @xmath42 distributions contain all the m sub - types combined , for stars across a large range of distances .
@xcite have shown that the flare frequency distribution slopes are similar , though not constant , for different spectral type bins , and between active and inactive stars @xcite and @xcite found that `` microflares , '' very frequent low - energy rapid flares , were detectable using hst for two very active m dwarfs . by comparing the distribution of fluxes about the median quiescent value in their light curves , the microflares , which are of too low amplitude to be detected individually , could be detected statistically .
they found that these microflare events followed a power - law distribution which was steeper than that of the larger resolved flares previously detected .
thus there are even greater numbers of microflares than would be predicted by simply extending the power - law flare frequency to lower energies .
our method of finding the residual @xmath42 distribution yields a similar result , by determining an asymmetry in the light curves via the @xmath42 statistic due to excess flux enhancements .
however we caution that we are not detecting the extremely low - amplitude microflares seen in the hst observations from @xcite . instead ,
the residual power in the @xmath42 distributions are likely a result of the underlying cacophony of low amplitude flares that are below our single - epoch detection limits in each bandpass , but which would be visible in high - precision continuous monitoring programs .
it is especially notable that the significant asymmetric distributions were seen in all wavelengths studied ( @xmath3@xmath4 ) , although we are sampling very different contrasts with the underlying quiescent stellar spectrum , which supports our interpretation of these events as flares .
while the flare frequency distributions in figure [ ffd ] are under sampled for the nir filters , the residual epochs with excess flux seen in the @xmath65 and @xmath86 in figure [ phi ] , and their power - law slopes in figure [ resid ] , provide a statistically significant detection of small - amplitude flux enhancements .
distributions , as defined in eqn [ phieqn ] , for each filter combination .
the steeper slope at increasing wavelength is due to the lower intrinsic rate of variability within the photometric errors.,width=336 ]
we have employed a semi - analytical two - component flare model , as first advocated by @xcite , to explore the completeness of our search for flares in every filter and to compare the observed flare luminosities between filters .
we extend the @xcite model which was used in blue optical wavelengths to red optical and nir bandpasses . in order to simulate flux enhancements due to flares in each filter using the two - component model , we built quiescent template m dwarf spectra in the optical and nir .
we utilized flux calibrated spectra from the irtf stellar catalog @xcite for stars with spectral type m0 through m6 .
the m0 through m6 sdss templates from @xcite were used for the most of the optical regime .
the irtf and bochanski spectra were normalized and joined at 8500 .
since the bochanski optical templates only go as short as 3825 , to fully cover the m dwarf spectra in the @xmath3-band we used templates from @xcite . as these are lower resolution than the bochanski templates , we used them only from 2900 to 5000 .
the pickles and bochanski templates are both published in normalized flux units , while the flux - calibrated irtf data we used in the nir are in observed flux .
we found distances to each of the irtf m dwarf standard stars from @xcite , @xcite , , and @xcite using the simbad database , and standard radii from @xcite to convert the observed flux into surface luminosity for each star in the nir .
the normalized optical templates were then scaled to match at 8500 , and or resultant optical / nir spectral energy distributions for the m0 through m6 sub types are shown in figure [ sed ] .
, for the @xmath3-band using four different @xmath42 thresholds to select flare candidate epochs .
lower thresholds allow more contaminating epochs into our flare distribution , but the power - law slope seen at luminosities above @xmath88 erg s@xmath70 remains the same.,width=336 ] the lower panel of figure [ model ] presents our m3 spectrum in arbitrary flux units in black , with the @xmath89 transmission curves from @xcite and @xcite shown for reference .
the overall flare sed shape was defined by a 10,000k blackbody continuum .
previous studies have found temperatures as low as @xmath08500k @xcite for flares , while @xcite found evidence for temperatures as high as 13,000k . fit a blackbody temperature of 20,000
k during the brief impulsive phase of an enormous flare on cn leo .
these temperatures rapidly declined , however , with values around 7,00010,000 k for the remainder of the flare event .
shortward of 3646 , we added a second component of h balmer continuum line emission , which was computed from the @xcite radiative hydrodynamic m dwarf atmosphere model .
the balmer continuum was scaled to have 10 times larger surface area coverage than the blackbody component .
the 10:1 scaling was approximately an average of those used by previous studies , with @xcite using values from 3:1 to 16:1 .
our idealized model does not include many detailed emission mechanisms that would alter these fluxes , such as line emission from other elements , or other levels of hydrogen .
using measured spectral line emission from m dwarf flares in the nir from @xcite , we estimate that line emission would contribute an order of magnitude less than our predicted blackbody continuum emission in @xmath34 bands , and is therefore negligible for our analysis . continuum emission from higher levels of hydrogen in the flare atmosphere , and effects from chromospheric back - warming such as discussed in @xcite ,
likely provide the greatest changes to the red optical and nir flare sed . to investigate the effect of the flare model , we scale it by a surface coverage fraction , ranging from 5@xmath90 to 0.5 , and add it to the m0m6 template spectra . the photometric response to flares in each filter was then calculated by convolving the sdss and 2mass filter transmission curves @xcite to the m0m6 model spectra with and without the flare . in figure [ dmag ]
we show the transformations between the @xmath91-band and the @xmath92 band amplitudes for all the flare models we ran .
m spectral energy distributions for our m0-m6 templates .
the ranges in wavelength over which each of the template sources contributes are shown at top.,width=336 ] the top panel of figure [ model ] shows the changes in each photometric bandpass as computed by our model for a flare on an m3 star with 1% surface coverage for the balmer continuum and 0.1% coverage for the blackbody component .
the resulting increases in flux for the @xmath89 bands were approximately ( 2.7 , 0.62 , 0.19 , 0.06 , 0.03 , 0.01 , 0.01 , 0.01 ) magnitudes respectively .
the extremely blue response of the flare is due to a contrast effect between the intrinsically blue flare model and the underlying red stellar spectrum .
since the m3 spectrum is much redder than the flare model , even large amplitude flares do not give a strong singal in the reddest filters .
however , k09 recovered a very bright flare in the stripe 82 sample , visible in all five passband light curves , on an active m6 star .
this flare had magnitude excursions of ( 5.50 , 3.07 , 2.34 , 0.77 , 0.29 ) for the @xmath39 bands respectively .
our m6 model , scaled to match the @xmath3-band excursion , predicted @xmath39 magnitude excursions of ( 5.5 , 2.5 , 1.2 , 0.3 , 0.1 ) . in the 2mass @xmath34 filters we predicted excursions of approximately ( 0.03 , 0.01 , 0.01 ) .
the coverage for this simulated flare was 0.06% of the stellar surface .
as the two - component model does not include detailed effects , as discussed above , the model excursions agree with the observations to within a factor of two in flux .
the k09 flare may also have evolved in temperature over the course of the sdss observations .
we allowed the blackbody temperature within our model to vary from 8,000k to 13,000k .
lower temperatures yielded only a marginally improved fit for the @xmath85-band , and required a larger effective surface area for the flare .
-band response and the @xmath93-band response for each spectral type template.,width=350 ]
using our flare model with the m3 spectral template , we investigated the efficiency of recovering flares from our sample with the @xmath42 statistic .
we first computed the sigma - clipped mean photometric error profiles as a function of observed magnitude for each of the @xmath89 bands in increments of 0.1 mag , shown in figure [ meanerror ] . for each of the m3 stars in our nir time domain sample
, we simulated 200 artificial flares using our model from 5 , with a logarithmically increasing surface area coverage fraction from @xmath94 to @xmath95 .
the mean error profiles were then used to select an appropriate photometric error for the resultant apparent magnitudes of each simulated flare in all eight passbands . finally , we calculated the @xmath42 values for each simulated flare on every star .
the result was a `` @xmath42 profile '' , which was a function of both the quiescent apparent magnitude of the star , and the flare amplitude .
these represent the two competing physical quantities , the distance to the star and the size of the flare , which conspire to limit our detection of small flares or flares on distant stars .
this envelope is shown in figure [ response ] as a function of quiescent magnitude on the x - axis , with @xmath42 contour values increasing from black to red .
the @xmath42 profile was smoothed with a kernel of 0.1 magnitudes . because flares are typically characterized by their enhancement of blue flux ,
the corresponding predicted @xmath3-band excursion from our model is shown on the right axis in blue for the @xmath93 bands . using our described @xmath42 thresholds from 4
, we found that small flares in the @xmath3-band ( @xmath96mag ) would be recovered for every star in the sample with a quiescent magnitude of @xmath97 .
the completeness is a direct function of the photometric error profile , which parallels the @xmath42 threshold curves shown in figure [ response ] .
interestingly , note that in the @xmath98 bands in figure [ response ] , the @xmath42 profiles `` turn up '' for bright stars , since the error profiles begin to increase as stars approach the saturation limit for the sdss . consider a star with a moderate quiescent luminosity in the @xmath3-band ( e.g. @xmath99 ) .
a very large flare such those observed in @xcite and @xcite would result in an observed @xmath3-band magnitude too bright for the sdss photometry , and would therefore be excluded from our analysis
. this would be possible in the @xmath11 , and even @xmath85-band , for bright stars with very large flares .
incompleteness is therefore also expected at the brightest flare luminosities , although these events are rare with duty cycles estimated to be @xmath01 per month on most active stars @xcite .
the error profiles in the @xmath11 , @xmath85 , and @xmath100 filters plotted in figure [ meanerror ] are best suited to monitor small amplitude variations such as eclipses and spot modulation .
figures [ response ] and [ dmag ] clearly show that we are only completely sampling flares larger than @xmath101 in these bands for most of our sample , and thus the @xmath3-band is the most suitable to search for the frequent low - amplitude flares in these stars .
as expected from the model discussion above , only very large flares would be detectable in the nir passbands for the brightest stars in our sample using the 2mass photometric error profile .
our model predicts that a flare with a nearly @xmath102 magnitude increase in the @xmath3-band would be required to detect any significant increase in flux in the cal - pswdb for a nearby m3 type star .
no simulated flares on m3 stars showed enough increase in flux to register as a @xmath103 response in the @xmath34 bands , however .
the flare emission in the nir must therefore be greatly influenced by physics not included in our two - component model , and definitive simultaneous detection of flares in the nir and optical is needed to calibrate this model .
our analysis has extrapolated a well - known two - component flare model into a new wavelength regime , and we therefore provide a prediction for the `` best - case '' to detect m dwarf flares in the nir .
an enhancement of @xmath104 magnitudes is easily observable for many nearby m dwarfs with present day ground based differential photometry .
using the @xmath91 to @xmath105 flare amplitude transformations shown in figure [ dmag ] , we estimated the @xmath91 enhancement for each spectral type required to achieve @xmath106 .
for spectral types m m6
this yielded a @xmath107(1.4 , 1.8 , 2.0 , 2.5 , 3.3 , 4.0 , 4.5 ) mag .
the changing shape in the underlying stellar sed in the nir indicates that @xmath7-band flares are detectable from smaller @xmath91 flares at earlier spectral types .
the frequency of flaring has been previously found to increase with spectral type , such as in our figure [ ffrac ] .
the characteristic lifetime of magnetic activity is also highly correlated with spectral type @xcite , and later spectral types m dwarfs are expected to show not only more , but larger amplitude flares ( hilton 2011 phd thesis ) .
this highlights the contrasting effects at work in detecting broadband nir flares : persistently active m0 stars that show large amplitude flares are exceedingly rare , as compared to those at later spectral types .
the higher bolometric luminosity of earlier spectral type stars additionally means much larger energy flares are needed to achieve these large amplitudes .
figure [ ffd_sptype ] shows the observed @xmath3-band flare frequency distribution for each spectral type bin as a function of @xmath91 magnitude .
the total rate of flaring increased with spectral type , and the slopes were very similar for all the distributions , in agreement with results from e. j. hilton ( 2011 phd thesis ) .
a linear fit was found for each log@xmath108 distribution between @xmath109 to avoid false - positives at the low - amplitude end , and incompleteness at high @xmath91 amplitudes . the predicted frequency at which we expect to observe a @xmath110 mag flare , from the corresponding @xmath91 amplitudes , at each spectral type
was then calculated from these fits , and is shown in figure [ rate_sptype ] .
the optimal spectral type to search for these nir flare detections is in our latest type stars , m6 .
the slight up - turn at m0 is due to the more favorable color - contrast between the flare and the star .
the overall trend from m1m6 , however , shows that the strongest effect for all spectral types is the total flaring duty cycle .
this provides one possible reason that the recovered flare rates are higher than expected for m0m1 stars in figure [ ffrac ] .
a new method for calculating the duty cycle of a given amplitude excursion , known as the magnitude likelihood distribution , has been recently put forward by e. j. hilton ( 2011 phd thesis ) . using these magnitude likelihood distributions
we predict a peak duty cycle of @xmath111 for the magnetically active m4 and m5 stars .
for the 54 sec exposures of sdss , this is equal to a frequency of @xmath112 hr@xmath70 , which is consistent with our estimation .
we would encourage future nir flare observers to monitor bright / nearby targets in the latest spectral types possible , at least m4 or later .
promising stars might include gj 1061 , cn leo ( wolf 359 ) , and especially the highly active yz cmi .
-band flares at each spectral type .
each distribution was fit with a linear polynomial in the range @xmath113.,width=336 ]
we have shown that the signatures of flares from m dwarfs are not confined to the bluest bandpasses in large scale photometric surveys .
there is significant response from medium to large flares in @xmath114 bands , and the largest flares are observable in @xmath115 and in some cases into the nir passbands .
this is the first statistical characterization of m dwarf flares in the nir regime .
a self - consistent result between all eight filters shows that while the frequency of flares decreases with increasing luminosity in all filters , the total number of flares recovered decreases with wavelength .
this suggests that like our simplified flare model the underlying sed for flares of all sizes is intrinsically blue .
furthermore , the total rate of flares in all optical bands appears to increase with increasing spectral type , but low - number statistics prevent us from robustly exploring this distribution at nir wavelengths .
the slope of the flare frequency distribution is found to be insensitive to the flare detection threshold below @xmath88 erg s@xmath70 .
mag amplitude flare is expected to be seen on each spectral type , based on our model and measured flare rates .
the numbers below each point are the @xmath91 amplitudes that our flare model predicts from a @xmath110 mag flare at each spectral type.,width=336 ] we have also presented an implementation of the two - component spectral flare model from @xcite to make predictions of the red - optical and nir response of flares in m dwarfs
. these conversions will be useful in future broadband photometric studies of m dwarfs to predict the contamination from flares across a wider wavelength space .
flares in the @xmath7-band are predicted to be detected most frequently in spectral types from m4 to m6 , with rates of @xmath116 hr@xmath70 .
medium sized flares larger than 0.5 magnitudes in @xmath3 were found to occur at a rate of one per few thousand sdss exposures for m dwarfs .
low - amplitude flare rates are limited by the photometric precision of the survey in question , and lsst will detect even smaller amplitude flares from these stars , and will increase the rate at which variability from these sources is detected .
lsst will also employ a @xmath117-filter , with a central wavelength between that of @xmath115 and @xmath7 . using the `` @xmath118 '' filter curve for lsst ,
our two - component flare model from figure [ model ] predicts a flare amplitude of @xmath119 mag .
the increased photometric sensitivity will therefore allow lsst to routinely detect large flares in this nir bandpass for bright stars , providing useful constraints on the sed structure of flares at long wavelengths finally , we have shown that the intrinsic sed of a flare , convolved with the photometric characteristics of a given survey , can greatly affect the sample of flares recovered .
the cadence and integration times will also affect the recovery of flares .
short exposures such as in 2mass capture the near instantaneous luminosity of an event , while longer exposure times may encapsulate entire flare events .
we anticipate a day when this problem can be inverted , using orders of magnitude larger samples of flare candidate epochs in many bands , as well as spectra , from next - generation surveys to probe the underlying flare sed ( or its deviations ) across the population of m dwarfs in the galaxy .
the authors would like to thank c. macleod and j. ruan for their insight in characterizing astrophysical variability , b. tofflemire and j. wisniewski for fruitful discussions regarding nir flares , and p. kundurthy for his assistance in converting times to bjd .
acb and jrad acknowledge support from nasa adp grant nnx09ac77 g .
ejh , slh , and afk acknowledge support from nsf grant ast 08 - 07205 .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center , california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
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 , university of cambridge , 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 .
ankerst , m. , breunig , m. m. , kriegel , h .-
p . , & sander , j. 1999 , in sigmod 1999 , proceedings acm sigmod international conference on management of data , june 1 - 3 , 1999 , philadephia , pennsylvania , usa , ed .
a. delis , c. faloutsos , & s. ghandeharizadeh ( acm press ) , 4960 | we present the first rates of flares from m dwarf stars in both red optical and near infrared ( nir ) filters .
we have studied @xmath050,000 m dwarfs from the sdss stripe 82 area , and 1,321 m dwarfs from the 2mass calibration scan point source working database that overlap sdss imaging fields .
we assign photometric spectral types from m0 to m6 using @xmath1 and @xmath2 colors for every star in our sample .
stripe 82 stars each have 50 - 100 epochs of data , while 2mass calibration stars have @xmath01900 epochs . from these data we estimate the observed rates and theoretical detection thresholds for flares in eight photometric bands as a function of spectral type .
optical flare rates are found to be in agreement with previous studies , while the frequency per hour of nir flare detections is found to be more than two orders of magnitude lower .
an excess of small amplitude flux increases in all bands exhibits a power - law distribution , which we interpret as the result of flares below our detection thresholds . in order to investigate the recovery efficiency for flares in each filter ,
we extend a two - component flare model into the nir . quiescent m0m6 spectral templates were used with the model to predict the photometric response of flares from @xmath3 to @xmath4 .
we determine that red optical filters are sensitive to flares with @xmath3-band amplitudes @xmath52 mag , and nir filters to flares with @xmath64.5 mag .
our model predicts that m0 stars have the best color - contrast for @xmath7-band detections , but m4m6 stars should show the highest rate of nir flares with amplitudes of @xmath80.01 mag . characterizing flare rates and photometric variations at longer wavelengths is important for predicting the signatures of m dwarf variability in next - generation surveys , and we discuss their impact on surveys such as lsst . |
stellar systems are conventionally divided into two broad categories , stellar clusters and galaxies , depending on whether they lie within a dark matter halo .
yet this scheme faces conceptual and practical difficulties .
for example , the dynamical masses of certain galaxies " ( ultra - compact dwarf galaxies or ucds ; * ? ? ?
* ) suggest little or no dark matter , as does the modeling of tidal dwarfs @xcite .
furthermore , even for indisputable galaxies like normal ellipticals , it can be difficult to infer dark matter from kinematic measurements within the luminous radius @xcite .
it is the spatial concentration of baryons relative to dark matter within the luminous region that distinguishes ucds from other galaxies and that makes dark matter easier to measure in spirals than ellipticals . rather than differentiating stellar systems with a binary classification
dark matter halo vs. no halo that is difficult to measure , we consider baryon vs. dark matter concentration , which is easier to observe and allows for a continuum of properties .
our previous work on scaling relations @xcite shows that galaxies lie on a progression of star formation efficiencies and baryon concentrations within dark matter halos .
the mass - to - light ratio @xmath0 within the half - light radius @xmath1 is one gauge of the relative concentration of baryons to dark matter .
the large @xmath0 s of dwarf spheroidals ( dsphs ) reflect the importance of dark matter within their luminous region , while the lower @xmath0 s of ellipticals arise from the dominance of the baryons there .
one could hypothesize a more extreme case than that for ellipticals , where the baryons are packed so tightly that the dark matter is dynamically negligible within @xmath1 .
it would be difficult to differentiate such a low @xmath0 galaxy , if it were low mass , from a star cluster , which is devoid of dark matter .
if they have any dark matter , ucds @xcite may be examples of galaxies whose spatial and kinematic structure thus resemble those of star clusters .
are there other observables that reveal their presumably distinct formation and evolutionary paths ?
remarkably , comparably low mass systems can have low @xmath0 ( stellar clusters , ucds ) or high @xmath0 ( dsphs ) .
what variations in dark halo properties of baryonic processes explain this wide range of baryon to dark matter concentrations ?
such questions are fundamental to understanding the nature of low mass stellar systems , the basic characteristics of galaxies , and the ways in which baryons can settle into a dark matter halo . here
we use a new scaling relation ( the fundamental manifold of ( fm ; * ? ? ?
* ) ) to explore the first question above , whether star clusters can be distinguished from low mass galaxies with similar @xmath0 s and the implications for their evolution . in certain projections of parameter space ,
globular clusters are distinct from galaxies @xcite .
( then again , in certain projections , dwarf galaxies are distinct from giant galaxies . )
projections of distributions in multi - dimensional parameter space can be misleading .
it is critical that the defined parameter space 1 ) has physically meaningful axes , so that one can interpret the results , 2 ) produces a galaxy distribution as thin as that allowed by observational errors ( i.e. , that has @xmath2 dimensionality ) , thus ensuring that the full dimensionality of the distribution is captured , and 3 ) exploits the full range of galaxy properties to maximize the discriminatory power of the observational space . in other words ,
we seek the minimum set of physically - meaningful parameters that captures the majority of the variance in galaxy properties .
the fundamental manifold , so named in reference to its antecedent , the fundamental plane @xcite , is tightly populated by galaxies of all morphological types and luminosities . by asking where star clusters lie relative to this manifold
, we have a means to quantify how distinct star clusters are from galaxies .
the relationship is straightforward @xmath3 where @xmath4 is the half - light radius in kpc , @xmath5 represents the internal velocity of the system and is a combination of the velocity dispersion , @xmath6 , and the rotational velocity , @xmath7 , in km s@xmath8 , @xmath9 is the mean surface brightness within @xmath4 in solar luminosities per square parsec , @xmath0 is the total mass - to - light ratio of the galaxy within @xmath4 in solar units , and @xmath10 is a normalization constant . additionally , and what leads to the finding that the systems populate a 2-d manifold , is the empirical determination that @xmath0 is primarily a function of @xmath5 and @xmath9 .
the functional form of @xmath11 is not simple , hence the broad term manifold , and while the data used to derive the fitted surface extend over many orders of magnitude along each of the relevant parameters , the function is unconstrained over much of the plausibly accessible parameter space and must be used with caution .
for example , the function as described by @xcite was unconstrained in the region of parameter space occupied by star clusters .
equation ( 1 ) originates from the virial theorem , and when combined with the finding that @xmath12 , it limits the region of parameter space occupied by the galaxy population .
in other words , _ the virial theorem alone allows infinitely more possible configurations that those found to populate the fm . _ as such , the fm places a strong constraint on galaxy formation and evolution models .
the simplicity of equation ( 1 ) and its relation to the virial theorem satisfies our first condition above .
the empirical finding that the scatter about this relation is small for the most diverse set of galaxies available , and yet captures their variety in kinematics , size , and luminosity , demonstrates that we have included the full dimensionality of the family of galaxies , at least to within the observed scatter , and therefore satisfies our second and third conditions . before testing whether star clusters lie on the fm , it is important to understand what the relationship implies about galaxies .
first , satisfying equation ( 1 ) requires that galaxies be in virial equilibrium and that they have what we will refer to as structural terms " that have a small degree of scatter .
hereafter , we use the term structure to encompass both the spatial distribution and kinematics of stars and dark matter .
these structural terms are the calculated numerical coefficients of the @xmath13 and @xmath14 terms one would obtain when evaluating the kinetic or potential energy terms in the virial theorem .
for example , an imaginary galaxy of constant density out to @xmath4 would have a different total potential energy than a galaxy that consists of the same amount of mass in a ring of radius @xmath4 , although the potential energy of either galaxy can be expressed as being proportional to @xmath15 .
it is impossible to numerically tally the terms in the virial theorem because one never has complete information regarding the phase space distributions of the stars and dark matter in a galaxy .
nevertheless , the low scatter and lack of systematic differences among different galaxy types when applying equation ( 1 ) indicates that the structural terms , which we have blithely incorporated into the constant term , are not detectably different among galaxies .
second , and perhaps somewhat more surprising , is that @xmath0 is principally determined by @xmath5 and @xmath9 , and therefore only modestly related to other factors that might have had an influence on @xmath0 , beyond whatever joint influence those factors may have had on @xmath5 and @xmath9 .
as such , there is a strong connection between the overall structure of a galaxy and the mass - to - light ratio within @xmath4 .
it is this second part in particular , the interdependence of @xmath5 , @xmath9 and @xmath0 , that one might suspect would be qualitatively distinct for star clusters if their formation or evolution mechanism(s ) is wholly different than that of galaxies .
consider that dsphs , with their @xmath16 velocity dispersions , are quite different in appearance from star clusters with similarly low velocity dispersion .
our plan is to test whether star clusters can be incorporated easily into the fm construct .
if they can be , then they can be described as a low - mass , high - concentration extension of the family of galaxies because the lack of dark matter within @xmath4 has produced no detectable difference in their structure relative to galaxies . on the other hand ,
if clusters lie significantly off the fm then their structure is sufficiently different from that of galaxies to suggest differences in their formation and/or evolution . to provide a specific example
, one could imagine a system like a star cluster that is instead dominated by dark matter ( consider turning some fraction of the stars into dark matter ) .
this system would have the kinematics and size typical of normal star clusters , but a very different surface brightness and mass - to - light ratio .
if we then allow the dark matter to follow a different radial density profile , all the observables would be affected and we would not necessarily expect this system to lie on the same relationship between @xmath0 , @xmath5 , and @xmath9 .
the converse does not hold in the sense that systems on the fm do not necessarily have the same evolutionary history .
although a better understanding of the origin of galaxies and star clusters is the primary goal , universal scaling relationships provide numerous ancillary benefits .
for example , we are familiar with the distance determinations using the fundamental plane ( fp ) and tully - fisher ( tf ) relationships .
unlike those scaling relations , the fm also provides a measure of the total mass - to - light ratio within @xmath4 .
given that star clusters lack dark matter halos , we will use the fm to constrain the _
stellar _ mass - to - light ratio , @xmath17 , as a function of cluster parameters such as main sequence turn - off age and color .
we will then compare these measures to values calculated using stellar population models as a test of those models . in summary , by considering star clusters and ucds , we test whether the fm formalism extends to the low - mass , stellar - dominated extremum of gravitationally - bound stellar systems in 3 .
we find that star clusters do lie on the fm but with apparently greater scatter than observed among the galaxies .
we then explore both observational and physical sources of that scatter .
we find that there is a correlation between cluster age and deviation from the fm , in the sense expected for variations in @xmath0 due to the evolution of the stellar population , but in quantitative disagreement with the predictions from stellar population models .
we explore this behavior and present empirical derivations of the relationships between @xmath0 , age , and @xmath18 in 4 .
we find that the dominant source of scatter in the current data set is observational , arising from the uncertainties in the cluster velocity dispersions .
we use the fm to make predictions for the velocity dispersions of those clusters without measured dispersions . in
5 we summarize our findings and discuss the implications . in an appendix we revisit previous claims that are in apparent conflict with our results @xcite and provide a resolution .
we draw the measurements of structural parameters , velocity dispersions , _ modeled _ stellar population mass - to - light ratios ( @xmath19 ) , ages , colors ( @xmath18 ) , and metallicities ( @xmath20 ) from the compilation by @xcite of local group star clusters .
that sample consists of 153 spatially resolved clusters from a range of nearby galaxies . within that sample
there is a subset of 57 clusters for which they have also compiled velocity dispersion measurements . as such
, it includes clusters of all ages , a wide range of metallicities , and the full range of environments found in the local group .
the data for stellar systems ranging from dwarf spheroidals to galaxy cluster spheroids ( csphs to distinguish them from stellar clusters throughout , see * ? ? ?
* ) are drawn from @xcite and for ultra - compact dwarfs from @xcite , selecting only objects with estimated masses @xmath21 to distinguish the ucds from the stellar clusters in their sample .
to define the manifold , we use only the spheroidal galaxies and csphs from @xcite as our comparison sample to minimize potential problems arising from composite stellar populations and to limit ourselves to systems supported by velocity dispersion rather than rotation . in the vast majority of local galaxies ,
either @xmath7 or @xmath6 dominates , and therefore we have not yet truly tested whether @xmath22 is preferable to other parameterizations , such as @xmath23 $ ] , or whether 0.5 is the optimal coefficient for the @xmath7 term ( this topic will be addressed in another paper , but values of 0.3 to 0.5 are plausible in idealized models ; * ? ? ?
* ) and such parameterizations have been used elsewhere @xcite . here , we avoid questions regarding the incorporation of rotation supported systems in the fm formalism by only analyzing systems in which @xmath6 dominates . the low relative rotation of milky way clusters is inferred from their low ellipticities @xcite . this argument is weaker for other cluster populations , such as those in the magellanic clouds , where larger ellipticities are sometimes found . however , even in the clouds , the cluster ellipticities are generally less than 0.5 @xcite , similar to the ellipticity of one of the most extreme mw clusters , @xmath24 cen , which has @xmath25 @xcite .
therefore , for the remainder of this study , we adopt @xmath26 . using only spheroidal galaxies
further simplifies our work by removing systems with ongoing or recent star formation , for which @xmath0 may have a dramatic dependence on mean stellar age .
the final relevant technical issue with the galaxy sample is the heterogeneous nature of the source samples and how one should compare it to the cluster sample . in particular
, the cluster sample contains @xmath5-band photometry , while the galaxy sample contains a range of photometry from @xmath5 to @xmath27-bands .
equation ( 1 ) is band - independent in that @xmath9 and @xmath0 for a particular object need to be in the same band , but that band can be different for different objects .
however , the manifold that we fit to describe @xmath0 , @xmath28 , is band - specific and has been derived using the heterogeneous galaxy sample , corrected to the @xmath27-band .
the values of @xmath9 used to determine @xmath28 are in solar units , so the filter dependence arises from the difference in colors between the galaxy and the sun when converting @xmath29 to @xmath9 . in effect
we are assuming solar colors for the galaxies , although any uniform color difference among all galaxies is absorbed into the calibration of @xmath10 in equation ( 1 ) .
the remaining concerns are color differences that vary across the sample and band - related differences in @xmath4 .
if these exist , then they are inappropriately incorporated into our fit for @xmath0 .
one clear path to progress is obtaining a single , uniform data set for the further study of the fm .
however , in the interim , we proceed with the motley sample available and caution that for systems with colors much different than solar colors there will be a yet unknown color - correction to @xmath28 .
the empirically - confirmed small scatter among galaxies , even when including star forming galaxies @xcite , suggests that this effect will be negligible for a sample consisting only of spheroidal galaxies , but could be a factor among the star clusters , which span a wider age range .
we revaluate @xmath28 using only those galaxies , the spheroids , from @xcite that are @xmath6 dominated , including galaxy cluster spheroids ( csph ) . in doing so
, we have also reevaluated the constant @xmath10 in equation ( 1 ) using the @xcite sample , for which they have full dynamical modeling fit to 2-dimensional spectral data for spheroidal galaxies , and find @xmath30 . in figure
[ fig : fit ] we show where the 57 globular clusters with measured velocity dispersions and the ucds with @xmath31 from @xcite fall when using this fitting function , _ which was not derived using either the star clusters or ucds_. in addition , we show with small points the range of possible locations in this projection of the extrapolated relationship covered by random , @xmath32 deviations of each of the fitted coefficients . the star clusters and ucds lie within the extrapolated version of @xmath28 given the fitting uncertainties . because the extrapolated fitting function works well in the mean for stellar clusters , we infer that clusters are indeed the low - mass , limiting case of high - stellar - mass - fraction systems and refit @xmath28 using the cluster and ucd data ( figure [ fig : fit1 ] ) .
the resulting fit is @xmath33 and is used hereafter .
projections of this function along the various axes ( figure [ fig : proj ] ) illustrate how difficult it is to see the multidimensional nature of the distribution and how certain populations of objects appear to stand out as different despite populating the same surface . because there is no physical motivation for this particular form
, the addition of new data , particularly those that probe a new part of parameter space , could require changing the functional form . however , the implications of the fm are independent of the particular functional form , depending only on the finding that @xmath28 is principally a function of @xmath5 and @xmath9
. the apportionment of the constant term between @xmath28 and the @xmath10 in equation ( 1 ) is set by the external calibration to independently measured values of @xmath0 and by construction in equation ( 1 ) that sets the two sides to be equal .
the scatter about the fm is evidently larger for the star clusters than it is for the galaxies ( 0.35 vs. @xmath34 0.1 ; * ? ? ?
* ) , so we turn to discussing possible sources of that scatter .
there are various potential causes of the larger scatter , including structural variations among clusters that are not present in galaxies , proportionally larger uncertainties in at least one of the observed parameters for the clusters , and the extrapolation of a fitted function beyond the parameter range over which it is constrained . among the observed parameters ,
the likely source of scatter is @xmath5 , which is more poorly measured in these low - velocity systems than in normal galaxies , and so we will begin our discussion by assessing the impact of the quoted uncertainties in @xmath5
. however , of greater interest is the possibility that some of the scatter is driven by a parameter that has not been included in our expression for @xmath28 .
this type of effect is often refereed to as a second parameter " question in correlations , although in our case it would be a third parameter .
if there is such a parameter , it has been inadvertently neglected in our formulation of the fitting function for @xmath28 . as such , @xmath35 as evaluated using equation ( 1 ) , @xmath36 ,
should correlate with this yet unidentified parameter ( as it does with @xmath5 and @xmath9 ) . in closing
, we note that there are two studies @xcite that appear to conflict with our claim that stellar clusters satisfy galaxy scaling relations .
we describe our interpretation of those studies in appendix a and conclude that neither study is truly in conflict with the results presented here . for these low velocity dispersion systems , the uncertainties in @xmath6
are fractionally the largest of all the input parameters .
we now explore the level of scatter about the fm expected simply due to the quoted uncertainties .
we use the 51 old ( @xmath37 gyr ) clusters with published velocity dispersion measurements and associated uncertainties to avoid any possible effect of age on the scatter .
we place each cluster exactly on the fm by adopting a value for @xmath0 that satisfies equation ( 1 ) .
we then draw a thousand cluster samples using gaussian distributed errors on the velocity dispersion , using the corresponding value for the uncertainty in @xmath6 for each particular cluster . in figure
[ fig : scatter ] we plot the distribution of the simulated scatter for the thousand trials and the observed scatter for comparison .
the observed scatter is somewhat larger than the typical simulated one , but we can not exclude the possibility that the observed scatter is directly due to the @xmath6 uncertainties with greater than 80% confidence . we conclude that the observed scatter in figure [ fig : fit ] is principally due to the uncertainties in @xmath5 , but can not conclude that no physical sources of scatter remain .
we test for physical sources of scatter in the next three subsections .
there is at least one physical cause of scatter that must play a role , that we can quantify easily , and that we can model variations in the stellar population mass - to - light ratio , @xmath17 . in figure
[ fig : fm - age ] we demonstrate that the observed deviations from the fm for star clusters are , as expected , different for the young and old stellar systems .
we fit a linear relation in @xmath38 to describe the residual ( @xmath39 and subsequently use this fit and the cluster ages from @xcite , obtained from resolved stellar main sequences , to correct @xmath28 for this systematic age deviation .
importantly , the residual is small for the older clusters , for which the calibration to galaxies is expected to be most appropriate .
there remain , however , at least two open questions .
first , the scatter remains large for both young or old clusters ( although , at least for the old clusters , we demonstrated above that observational uncertainties can account for the scatter ) .
second , the mean trend only testifies to some age - related effect , not that this effect must be related to @xmath17 .
there are three reasons that the trend is likely due to the evolving stellar population rather than to age - dependent dynamical effects : 1 ) the stellar population must , to some degree , evolve and affect @xmath0 because @xmath0 is related to @xmath17 , 2 ) the fitting function works best ( i.e. the residual is @xmath34 zero ) for the oldest clusters , for which the calibration for spheroidal galaxies and their similar stellar populations should be most appropriate , and 3 ) dynamical effects , such as those involving two - body relaxation , should trend in the opposite direction with age , becoming more , rather than less , important in the older clusters .
unfortunately , further discussion is limited by the lack of clusters with measured @xmath6 across our entire age range , @xmath40 } ) < 10 $ ] .
although the entire @xcite cluster sample does contain clusters in this age range , we need an estimate of @xmath6 to include them in our discussion .
we next describe how we use the fm to obtain such estimates . because the right hand side of equation ( 1 ) can be written as only a function of @xmath5 and @xmath9 after adopting @xmath28 , we use the tabulated values of @xmath9 and @xmath4 for the clusters to solve for @xmath5 , which we equate with @xmath6 ( see 3 ) .
the drawback of this approach is that we have now assumed , on the basis of figure [ fig : fit ] , that both equation ( 1 ) and @xmath28 are applicable to clusters .
we account for the trend of @xmath28 with age shown in figure [ fig : fm - age ] using the linear fit , and then we solve the equation for @xmath5 .
we test this procedure on the subset of 57 clusters with measured @xmath6 in figure [ fig : sigtest ] .
the agreement is encouraging and we proceed to estimate @xmath6 for all 153 clusters .
the agreement within the uncertainties of the calculated and observed @xmath6 s in figure [ fig : sigtest ] demonstrates that much of the scatter among clusters on the fm must come from the uncertainties in their respective @xmath6 s , as we had surmised on other grounds . in other words
, we have not created a situation where we absorb errors in other quantities or intrinsic scatter into our derived values of @xmath6 . in all subsequent plots ,
we use the inferred @xmath6 s , but highlight those clusters with measured @xmath6 s using large open circles to illustrate the degree to which any trends are driven by those clusters with unconfirmed velocity dispersion estimates .
an interesting subset of clusters are those of lower surface brightness in the outskirts of the milky way . because of their distance and naturally low @xmath6 ,
they tend to not have @xmath6 measurements of high relative precision
. outside of the compilation of cluster properties by @xcite that we have used exclusively , there are measurements of @xmath6 for pal 5 @xcite and pal 14 @xcite .
using these values , we find that even these clusters fall on the fm relation . as a target for future observers
, we present the calculated values of @xmath6 necessary to place similarly unusual clusters on the fm ( table [ tb : oddballs ] ) .
these low dispersions illustrate why high precision spectroscopy and a detailed knowledge of the binary populations ( cf .
@xcite ) are needed to confirm the calculated @xmath0 s and to resolve the question of whether such systems lie on the fm .
we now return to investigate the inferred drift in @xmath0 vs. age by including all star clusters with or without measured @xmath6 s .
we plot the total mass - to - light ratio within @xmath4 evaluated using equation ( 1 ) , @xmath36 , vs. age ( figure [ fig : allages ] ) . for those clusters where we estimate @xmath6 , the values of @xmath35 are not independent of the measured age , and so trends should be viewed with caution and compared to those obtained from clusters with measured @xmath6 . because stellar clusters are dark matter free , @xmath41 .
we confirm that the value of @xmath0 necessary to place clusters on the fm is a simple function of age and , as seen in figure [ fig : allages ] , that the sense of the trend ( higher values of @xmath0 for older ages ) is as expected and matches the behavior determined from only the clusters with measured @xmath6 . because all but the lowest four of the clusters in figure [ fig : fit ] are old , the apparent scatter is not the result of age differences within the sample .
another potential source of scatter is variance related to the cluster structural properties .
star clusters can evolve dynamically due to their relatively short dynamical times ( in comparison to the dynamical times for galaxies ) and therefore should have larger variations in their structural terms , which reflect both the stellar kinematics and spatial distribution .
we examine whether structural differences among the clusters have been incorrectly excluded from equation ( 1 ) , or @xmath28 , by looking for correlations between @xmath36 and measurements of the structural state of the cluster . in a similar approach ,
@xcite found some evidence for correlations between the structural state of galactic clusters and their location relative to the fp .
figure [ fig : comp_dyn ] shows the relationships between @xmath36 and central potential depth , concentration , and galactocentric distance .
whatever trends there may be in the lower panels , they are clearly of much lower magnitude than that with age , which is evident as the bimodal distribution in the upper panels of the figure . we conclude that none of the measures of cluster structure that are currently available suggest a relationship between deviations from the fm and the dynamical state of the cluster beyond any directly related to age . because dynamical state depends on more than just age ,
we conclude that stellar population differences , which do depend only on age , are the primary driver of the trends we observe .
next , we search for a dependence of @xmath36 on cluster metallicity ( figure [ fig : fe - corr ] )
. there is no evident dependence of @xmath36 on metallicity below @xmath42 in the upper panel of the figure .
there may be an effect for higher metallicities , although that is where age differences become important as well . in the lower panel ,
we apply the age correction derived from figure [ fig : fm - age ] and find that there is little if any residual metallicity dependence .
again , the principal driver of the trends appears to be age .
on the basis of the correlation between scatter and age , and the agreement between the inferred sense of the effect and the qualitative expectations from stellar population models , we conclude that age is the dominant physical source of scatter .
however , our data do not allow us to eliminate all possible structural differences that might play a role , nor do they statistically demonstrate the superiority of age as a driving parameter over the other parameters now available .
the peculiarities of the sample , only six clusters with measured @xmath6 at age @xmath43 yrs and the remainder all at old ages , preclude application of statistical tests such as principal component analysis or partial correlation coefficients as a means to identify a dominant physical effect .
however , this shortcoming can be remedied with reasonable observational effort .
the variation in @xmath35 , which for stellar clusters is equivalent to @xmath17 , vs. age has a natural explanation in terms of stellar population evolution . in figure
[ fig : comp ] we compare estimates of @xmath0 based on the fm , @xmath36 , to those based on the model values of @xmath17 , @xmath19 , tabulated by @xcite using their preferred bruzual - charlot models with a chabrier initial mass function ( imf ) .
the sense of the behavior of @xmath0 with time is consistent for both @xmath36 and @xmath19 suggesting that we are indeed seeing the effects of stellar evolution on @xmath0 .
however , the quantitative agreement is poor , with @xmath36 overestimating @xmath19 at young ages and underestimating it at old ages .
the existence of a significant difference between @xmath19 and @xmath36 , particularly at old ages , is independent of the particular model or initial mass function , among those currently in general use .
we compare @xmath36 to the full range of @xmath19 presented by @xcite in figure [ fig : stpops ] .
we find similar results after comparing to 2007 bruzual - charlot models , which are not included in @xcite .
the magnitude of the differences between @xmath19 and @xmath36 is similar to that of the differences among the stellar population models themselves .
it is therefore possible that there are problems with the stellar population models at this level .
@xcite explore the uncertainty in stellar population models , the bulk of which comes from modeling intermediate and low - mass stars near the end of their lives where they become far more luminous but for short periods of time .
@xcite show that including the thermally - pulsating agb phase can increase the emission from an intermediate age galaxy by 1 magnitude in the @xmath45-band .
the level of the discrepancy suggested by our fm analysis is consistent with these potential problems in @xmath19 . even at the young ( low @xmath0 ) end of things , there are clear differences among the models , with those using the salpeter imf producing results more in line with those from the fm
. further discussion of the details of stellar population models is beyond the scope of this paper , but the fm provides a new constraint .
the fm - derived values of @xmath17 suggest that all currently popular stellar population model variants over - predict the stellar mass - to - light ratio of old populations ( see * ? ? ? * for an alternate approach that results in the same conclusion ) .
stellar population models that provide accurate stellar mass - to - light ratios are manifestly important for a variety of uses .
however , star clusters on the fm provide a new way to estimate @xmath0 for large sets of _
galaxies_. currently , the fm zero point ( @xmath10 in equation ( 1 ) ) is calibrated using the results for @xmath0 from detailed dynamical studies of several tens of early - type galaxies @xcite .
if we knew , from stellar models , the values of @xmath17 , and hence @xmath0 , for star clusters , we could use them instead to calibrate the fm .
once the fm is accurately calibrated , one can use it to solve for @xmath0 for any galaxy with measured @xmath5 , @xmath4 , and @xmath9 .
we apply the fm to derive empirical relationships between parameters such as main sequence turn - off age and color , and @xmath17 .
figure [ fig : independent ] shows our crude fits to the age-@xmath17 ( table [ tb : fitage ] ) and ( b - v)-@xmath17 ( table [ tb : fitbmv ] ; the fit is valid for @xmath46 ) relationships .
neither 1-d relationship appears satisfactory for the full range of clusters , and certain populations , such as young and metal - poor populations , are completely unconstrained .
more sophisticated and well - constrained fits will be possible once more data become available for intermediate age clusters , providing an alternative to @xmath17 estimates that depend on stellar populations models .
we find the following : \1 ) on average , star clusters fall along the extrapolation of the fundamental manifold ( fm ) defined by spheroidal galaxies .
their larger scatter is consistent with the proportionally larger uncertainties in their velocity dispersion measurements .
even extreme clusters such as pal 5 and 14 fall on the relationship when precise velocity dispersions are available .
\2 ) individual clusters are offset from the fm in proportion to their age : older clusters fall on the manifold and younger clusters do not .
unfortunately , the sample of clusters with measured velocity dispersions does not include many clusters with @xmath47 } ) < 10 $ ] , so we apply the fm relationship to infer @xmath6 and @xmath0 simultaneously for the remainder .
this procedure works faithfully for those with measured @xmath6 .
\4 ) the sense of the evolution of mass - to - light ratio with age is as predicted by stellar population synthesis models , but currently popular models fail to reproduce the estimates from the fm ( see also * ? ? ?
specifically , the models over - predict the mass - to - light ratio of old , metal - rich populations . we provide empirical formulae with which to estimate the stellar mass - to - light ratios of certain cluster populations derived from the fm - estimated mass - to - light ratios .
all stellar systems that we have examined in detail so far brightest cluster galaxies @xcite , various galaxy types and luminosity classes @xcite , and now star clusters satisfy a simple , 3-parameter ( 2-dimensional , but not planar ) scaling relation . for star clusters , deviations from the scaling law are age dependent , which is qualitatively consistent with the expectation for passive stellar evolution of the population .
quantitatively , however , this description fails .
we suggest , but have not definitively demonstrated , that this disagreement arises from systematic errors in the stellar population models that lead to an overestimation of the v - band mass - to - light ratio of old stellar populations .
the sense and magnitude of the effect are consistent with conclusions reached by others investigating stellar evolution models in detail @xcite and at the very least suggest that a factor of two uncertainty remains in modeling stellar populations . as stressed by others ( cf .
* ; * ? ? ?
* ) , this uncertainty propagates in complicated ways , particularly for composite populations of unknown age distributions , in analyses of galaxy evolution .
we conclude that there are no evident structural differences , in either stellar spatial distribution or kinematics , between galaxies and star clusters beyond those of scale and those captured in the mass - to - light ratios within @xmath4 .
star clusters appear to be baryon - dominated versions of highly - compact , low - mass galaxies .
as such , we are not surprised by the difficulties encountered by others in distinguishing clusters from ultra - compact galaxies .
if there are wholesale differences between the formation process and/or evolution of star clusters and galaxies , they do not manifest themselves directly in the gross physical properties .
we conclude that even if significant differences exist between the populations , due to the existence or lack of a dark matter halo , simple measurements will not reveal them .
projections of parameter space in which these populations are separate are deceptive and should not be taken to indicate that these objects are discrete families of gravitationally bound stellar systems .
although this lack of evident distinction is a set - back for attempts to differentiate the populations and probe potential differences in their formation and evolution , it is an advantage to those efforts to build a simple set of models for galaxies .
star clusters can provide an anchor for models of galaxies , empirically defining the stellar mass - to - light ratios or , alternatively , the zero point of the fm .
the latter would enable us to use the fm to determine the mass - to - light ratio within @xmath4 , to a level of precision comparable to that obtained with detailed dynamical modeling or strong gravitational lensing modeling , for any galaxy with known @xmath48 , and @xmath9 .
we return to the question of the defining characteristics of a galaxy . among the stellar systems that we study ,
we have not identified a criterion that we consider a discriminating feature . despite the popularity of the dark matter halo criterion
, there exist galaxies " that may not have a dark matter halo ( tidal dwarf galaxies ; * ? ? ? * ) or , at least , a dominant dark matter halo ( ultra - compact dwarf galaxies ; * ? ? ?
in fact , from state - of - the - art optical data alone , it is difficult to demonstrate that even giant ellipticals contain dark matter @xcite .
an alternative , and more easily observable , classification is the relative concentration of luminous baryons to dark matter . in such a scheme , systems composed solely of baryons are not distinct from those with dark halos and whose observed
kinematics are dominated by baryons .
while this scheme makes it harder to connect observations with dark matter simulations , the reality is that , for systems whose baryons are extremely concentrated , we can not determine whether a halo exists or what its properties might be .
our primary empirical result is that the existing description of the fundamental manifold @xcite is valid over an even larger parameter range than previously demonstrated and that now includes stellar clusters .
the existence of a fm for the entire family of stellar systems implies 1 ) that the objects are virialized and 2 ) that there is limited variation in the overall structure . while the former is unsurprising , the latter implicitly places strong limits on how the systems form and how the luminous baryons settle . by extending the validity of the fm , or any scaling relation , over a larger range of parameter space and a wider set of objects
, we are demonstrating that the physics that constrains the resulting baryon and dark matter distributions in stellar systems is more general than previously thought .
in other words , whereas one might not have been surprised by the implication of the fundamental plane that all ellipticals have related structural properties , it is certainly a puzzle to find that spirals , dwarf spheroidals , and now stellar clusters do too .
while the structure of one stellar system is not a simple scaling of another , there exists a transform function that relates the properties of stellar systems .
specifically , we have presented a fitting function for @xmath0 that is valid across all stellar systems .
the generality of this transform function implies that 1 ) the structure of one type of stellar system is not a distinct result of a process particular to that set of systems , such as violent relaxation for elliptical galaxies , but is instead the result of an interplay of all processes responsible for the generic settling of baryons in gravitational potential wells , 2 ) the existence of the transform function for @xmath0 , and hence the nature of baryon settling , is independent of whether the system is embedded within a dark matter halo , and 3 ) differences in initial conditions or stochastic events during the lifetimes of stellar systems do not invalidate the transform function . by including stellar clusters , which we presume are devoid of dark matter
, we demonstrate that the nature of the transform function for @xmath0 depends more on the baryon physics than the source of the gravitational potential . the presence or absence of dark matter imprints
no clear feature in the structure or kinematics of stellar clusters versus galaxies . by including stellar clusters , which we presume formed mostly in single , short - lived , dramatic fashion
, we demonstrate that the nature of the transform function for @xmath0 is not altered radically by differences in formation history .
the extension of the fm over a larger parameter space is in reality the extension of the statement that physical processes rather than initial conditions or subsequent stochastic events dominate the final outcome in the formation of stellar systems .
dz acknowledges financial support for this work from a nasa ltsa award nng05ge82 g and nsf grant ast-0307482 .
aiz acknowledges financial support from nasa ltsa award nag5 - 11108 and from nsf grant ast-0206084 .
dz and aiz thank mpia and nyu for their hospitality and support during visits where this paper was completed . finally
, we thank the anonymous referee , whose comments helped us clarify various arguments .
+ am1 & 0.6 & ... & ... + arp2 & 0.7 & ... & ... + pal3 & 1.2 & ... & ... + pal4 & 1.1 & ... & ... + pal5 & 0.7 & [email protected] & @xcite + pal12 & 0.6 & ... & ... + pal14 & 0.5 & 0.38 @xmath49 0.12 & @xcite + [ tb : oddballs ] various efforts have been made to establish a fundamental plane for globular clusters @xcite .
however , by necessity , this plane is not the same as that for ellipticals , whose fundamental plane ( fp ) requires @xmath51 .
@xmath0 for intermediate mass systems is already comparable to @xmath17 and can not physically decrease further for systems with comparable surface brightnesses but lower velocity dispersions , like globular clusters @xcite .
a recent study discussed a unification of all spheroidal " systems from globular clusters through elliptical galaxies @xcite . while prevailing generally , they found that dwarf spheroidals did not fit their general formalism and that the fm @xcite was not valid down to globular clusters .
we disagree with the latter statement and so we investigate the nature of the discrepancy .
we use their data , which they made publicly available with their paper , to remove any sample differences .
their claim regarding the fm is based on the nature of the empirical scaling relation for @xmath0 . in particular
, they note that the globular clusters do not follow the u "- shaped pattern described for @xmath0 by @xcite . because their data include @xmath45-band magnitudes , our fitting formula for @xmath0 , which is calibrated to @xmath27-band values of @xmath0 , is not optimal ( color differences become large ) .
instead we refit using their data and obtain a version of @xmath28 appropriate for the @xmath45-band .
the central issue of the fm is whether the family of stellar systems can be described with a two - parameter relationship .
the coefficients in the expansion of @xmath28 , or even the functional form , are not sacrosanct , and currently have no detailed physical justification .
the result of the new fit is shown in figure [ fig : forbes ] .
as @xcite stressed , the most salient feature of this figure is not that the data straddle the 1:1 line , because this arises by construction ( although it does imply that our functional form for @xmath28 is an acceptable one ) , but rather the low and uniform scatter across the full range of systems ( i.e. , clusters , dwarf galaxies , and giant galaxies ) .
in the right panel of figure [ fig : forbes ] , we explore the crux of the argument against the fm presented by @xcite .
in particular , they noted that star clusters , with low values of @xmath0 , do not follow the upward trend in @xmath0 established by local group dsph galaxies .
the difficulty with this argument is that it is based on a projection of the data onto the @xmath52 axes . in the right - hand plot
, we show the value of @xmath28 for each object in the @xcite sample . by construction , because we use the values on the manifold , _ there is zero scatter in this distribution _ , and yet in this projection there appears to be significant scatter .
in fact , the clusters , which populate the lower left branch , appear to be a distinct population from the dsphs , which lie directly above the clusters and extend upward along the @xmath0 axis . therefore ,
even points distributed on a surface with zero scatter can , in projection , appear to comprise different populations of objects because the surface is not uniformly populated . as such
, one can not conclude from plots such as that shown on the left of the figure that there is no single scaling relationship among the populations .
it is therefore critical to consider the behavior of @xmath28 in terms of both @xmath6 and @xmath9 .
we find that the fm fits the characteristics of both high and low surface brightness objects with low ( @xmath53 km sec@xmath8 ) velocity dispersions .
in particular , in our figure [ fig : fit ] the fm was derived excluding the star clusters and ucds , and yet both populations fall within the uncertainties of the extrapolated relationship .
the apparent disagreement with @xcite stems from their aims and approach .
their aim was to identify the parameter with the greatest influence in producing the full range of spheroid properties along the scaling relationship for spheroids .
they correctly focused on velocity dispersion as the dominant parameter .
they proceeded to derive a one - parameter scaling relationship that they could then tie to halo mass . because they use only one parameter , they are unable to simultaneously fit the range of stellar systems with low @xmath6 ( i.e. , dwarf spheroidals and star clusters / ucds ) . given their interest in tying stellar systems to dark matter halos , they naturally accepted the fit that was consistent with the dsphs , which are clearly embedded in dark matter halos , and argued that the clusters / ucds are a different family distinguished by their lack of dark matter halos .
that is certainly one viable interpretation given the state of the data for ucds .
however , we argue that 1 ) there is some evidence for ucds having dark matter @xcite , 2 ) at the high - mass end , the ucds overlap nicely with the low - mass end of the galaxy distribution ( see figure 1 ) , and 3 ) disk galaxies also provide evidence for a two - parameter scaling relation .
our differences are therefore ones of interpretation regarding whether it is appropriate to consider all stellar systems together or to distinguish among them .
our preference for a single descriptive formalism for all stellar systems leads us in one direction , their preference for a 1:1 association of stellar systems with dark halos leads them in another . | we explore whether global observed properties , specifically half - light radii , mean surface brightness , and integrated stellar kinematics , suffice to unambiguously differentiate galaxies from star clusters , which presumably formed differently and lack dark matter halos .
we find that star clusters lie on the galaxy scaling relationship referred to as the fundamental manifold ( fm ) , on the extension of a sequence of compact galaxies , and so conclude that there is no simple way to differentiate star clusters from ultra - compact galaxies . by extending the validity of the fm over a larger range of parameter space and a wider set of objects ,
we demonstrate that the physics that constrains the resulting baryon and dark matter distributions in stellar systems is more general than previously appreciated .
the generality of the fm implies 1 ) that the stellar spatial distribution and kinematics of one type of stellar system do not arise solely from a process particular to that set of systems , such as violent relaxation for elliptical galaxies , but are instead the result of an interplay of all processes responsible for the generic settling of baryons in gravitational potential wells , 2 ) that the physics of how baryons settle is independent of whether the system is embedded within a dark matter halo , and 3 ) that peculiar initial conditions at formation or stochastic events during evolution do not ultimately disturb the overall regularity of baryonic settling .
we also utilize the relatively simple nature of star clusters to relate deviations from the fm to the age of the stellar population and find that stellar population models systematically and significantly over predict the mass - to - light ratios of old , metal - rich clusters .
we present an empirical calibration of stellar population mass - to - light ratios with age and color .
finally , we use the fm to estimate velocity dispersions for the low - surface brightness , outer halo clusters that lack such measurements . |
galaxy clusters are the largest self - gravitating systems in the universe , residing at the intersection of large - scale filamentary structures of the cosmic web . according to the hierarchical structure formation scenario based on a cold dark matter ( cdm ) paradigm , mass accretion flows onto clusters are still ongoing and high - mass ratio cluster mergers , the so - called major mergers , sometimes occur .
major cluster mergers are among the most energetic events in the universe , releasing amounts of gravitational energy as large as @xmath2 . spatially resolved x - ray observation has revealed to us that this collision energy partly dissipates in the intracluster medium ( icm ) through shock heating and turbulence , yielding a complex icm brightness and thermal structure ( see , e.g. , * ? ? ? * and reference therein ) .
the dynamics of cluster collisions are however dominated by the cluster mass distribution , which can not be constrained from x - ray observations alone . due to the collisional nature of icm , cluster mergers
are indeed expected to generate some transient decoupling between spatial distributions of the cdm and hot x - ray emitting gas .
moreover , mass estimates relying on gas properties may be biased by merger - induced perturbations of the icm hydrostatic equilibrium ( h.e . ) within individual colliding clusters .
weak - lensing distortions of background galaxy images provide us with a unique opportunity to reconstruct the distribution of matter in clusters without any assumption of mass model and dynamical states ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and measure cluster masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , a weak - lensing analysis of merging clusters ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) is an observational breakthrough in measuring mass distribution , thereby providing complementary information to x - ray measurements . @xcite
conducted a systematic study of seven merging clusters , representing various merging stages and conditions , based on a joint weak - lensing , optical photometric , and x - ray analysis , and revealed that the mass and optical light of member galaxies are similarly distributed in merging clusters regardless of their merging stages , but the mass distribution in merging clusters is highly irregular , and is quite different from the icm distributions .
it indicates that the gas and mass evolutions under cluster mergers are different .
this feature is confirmed by weak - lensing studies of 30 clusters as a collaboration of `` local cluster substructure survey '' @xcite .
thus , a joint analysis of clusters ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* ) will yield a comprehensive and quantitative understanding of cluster merger physics involved in the structure formation .
located at a redshift of @xmath3 , a2163 is a rich , x - ray luminous and hot galaxy cluster showing various signatures of ongoing merger events , including irregular optical and x - ray morphologies ( see , e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and a prominent radio halo emission @xcite .
this cluster has been known to exhibit a clear spatial segregation between a bimodal galaxy distribution and a more centrally peaked icm morphology ( * ? ? ?
* hereafter m08 ) , while the projected distance separating its main mass and gas centroids has been measured as one of the largest in a sample of 38 clusters analyzed from both strong galaxy lensing and x - ray imaging @xcite .
as revealed from spectroscopic and photometric analyses in m08 , the galaxy distribution in a2163 can be separated into two components : the massive cluster a2163-a and its northern companion a2163-b , and the main cluster component itself showing a bimodal morphology with two brightest galaxies ( bcg1 and bcg2 ) . in a more recent analysis of _ chandra _ and _ xmm - newton _ data , ( * ? ? ? * hereafter b11 ) evidenced the westward motion of a cool core across the e - w elongated atmosphere of the main cluster , a2163-a .
located close to the second galaxy over - density , this gas bullet appears to have been spatially separated from its galaxy component as a result of high - velocity accretion . from gas brightness and temperature profile analysis performed in two opposite regions of the main cluster ,
b11 further showed that the icm has been adiabatically compressed behind this crossing `` bullet '' .
characterized by an exceptionally high icm temperature first measured in x - ray ( @xmath4 kev ; * ? ? ?
* ) then confirmed from its sunyaev - zeldovich ( sz ) distortion @xcite , a2163 has been suggested to be exceptionally massive from various x - ray and sz analyses ( see , e.g. * ? ? ?
* ; * ? ? ?
* ) via the mass scaling relation .
former weak - lensing analyses have been conducted using the very large telescope and cfht / mega - cam analyses of @xcite and @xcite .
these analyses , however , did not exclude merger galaxies in the background shear catalog , as demonstrated by other studies @xcite . as shown by @xcite , a contamination of member galaxies
significantly dilutes lensing distortion signals , mainly in the central region , thereby yielding biased estimations on cluster parameters . in this paper
, we conducted a weak - lensing analysis using subaru / suprime - cam and made a secure selection to avoid a contamination of member galaxies in the shear catalog , combining cfht / mega - cam data .
the structure of this paper is as follows : in section [ sec : data ] we briefly describe the weak - lensing analysis .
the projected distributions of mass , member galaxies and hot - gas are compared in section [ sec : map ] .
section [ sec:1d ] is devoted to a one - dimensional shear analysis to measure cluster mass . in sec
[ sec:2d ] , we perform , based on a two dimensional shear pattern , a multi - components fitting to measure three components revealed by m08 .
we compare weak lensing mass with dynamical and x - ray hydrostatic equilibrium masses in section [ sec : mass ] . in section [ sec : dis ] , we discuss quantitatively the physical process of making an offset between mass and gas distributions .
section [ sec : sum ] summarizes our results .
the cosmological parameters of @xmath5 mpc@xmath6 km s@xmath6 , @xmath7 , and @xmath8 are used in this paper .
given the cosmology , @xmath9 .
we retrieved @xmath10 image data from the subaru archival data ( smoka ) .
the data were reduced by the standard imaging process of using the reduction software for suprime - cam , sdfred @xcite , as described in @xcite and @xcite .
astrometry calibration was conducted by fitting the final stacked image with the two micron all sky survey data catalog . the residual in astrometric fitting was lower than the ccd pixel size .
the exposure time is @xmath11 minutes and the seeing is @xmath12 ; therefore , it is suitable for our weak - lensing study .
since no other band data are available , we use cfht / mega - cam @xmath13 imaging data to exclude unlensed galaxies in the shear source catalog via a color - magnitude plane .
our weak - lensing analysis is done using the imcat package provided by n. kaiser ( kaiser et al .
we use the same pipeline as @xcite with some modifications followed by @xcite .
we first measure the image ellipticity , @xmath14 , from the weighted quadrupole moments of the surface brightness of each object and then correct the point - spread function ( psf ) anisotropy by solving @xmath15 , where @xmath16 is the smear polarizability tensor and the asterisk denotes the stellar objects .
we fit the stellar anisotropy kernel @xmath17 with the second - order bi - polynomials function in several subimages whose sizes are determined based on the typical coherent scale of the measured psf anisotropy pattern . the median stellar ellipticities before and after the anisotropic correction are @xmath18 and @xmath19 , respectively .
stellar ellipticities before and after the correction and their pattern are shown in figures [ fig : stell_e ] and [ fig : emap ] , respectively .
we next estimate the reduced shear @xmath20 using the pre - seeing shear polarizability tensor @xmath21 , where we adopt the scalar value @xmath22\delta_{\alpha\beta}/2 $ ] .
we select background galaxies in the range of @xmath23 , where @xmath24 is the half - light radius , and @xmath25 and @xmath26 are the median and standard error of stellar half - light radii , @xmath27 , corresponding to the half median width of circularized psf .
then , we make a secure selection of background galaxies in a color - magnitude plane in order to minimize a dilution of the weak - lensing signals caused by a contamination of unlensed galaxies , mainly member galaxies . as shown in @xcite , the dilution effect on the lensing signals
is more pronounced at smaller radii because the ratio of the number density of cluster galaxies to background ones rises towards the inner region .
we use subaru / suprime - cam @xmath10 and cfht / mega - cam @xmath13 images .
the difference in filter sensitivity functions of the wavelength in the @xmath13 band between these two instruments is negligible .
therefore , a combination with subaru and cfht enables us to efficiently exclude member galaxies inthe background shear catalog .
following @xcite , we calculate the lensing signal as a function of color ( figure [ fig : dilution ] ) , by averaging the tangential distortion strengths ( see also * ? ? ?
we securely select background galaxies in the range of @xmath28 , @xmath29 and @xmath30 , where @xmath31 is the best - fit linear function as a magnitude @xmath10 of red - sequence galaxies ; @xmath32 , as shown in figure [ fig : color - mag ] . here
, the magnitudes and colors are measured by mag_auto ( i.e. , total magnitude ) and mag_aper ( i.e. , aperture magnitude ) in sextractor , respectively . by a selection in the color magnitude plane , the number density of background galaxies in the shear catalog decreases from @xmath33 to @xmath34 .
the mean redshift of these background galaxies is estimated by matching the cosmos catalog @xcite to calculate the average lensing weight @xmath35 , where @xmath36 and @xmath37 are the angular diameter distances between the observer and source ( background galaxy ) and lens and source , respectively , and @xmath38 is a probability function of redshift distribution .
we obtain @xmath39 .
according to the cdm paradigm , the collisional icm is expected to experience a spatial decoupling from galaxies and dark matter during cluster collisions .
comparing hot x - ray emitting gas maps with the cluster member galaxy density and mass distributions has thus been known to testify to the merging cluster dynamical states ( see , e.g. * ? ? ?
* ; * ? ? ?
in order to investigate the collision scenario in a2163 , we map out the cluster mass distribution and compare it with spatial distributions of the galaxy and hot gas components .
as described in detail in @xcite , we pixelize the shear pattern into a regular grid using gaussian smoothing kernel @xmath40 $ ] with @xmath41 .
we adopt the smoothing scale of @xmath42 due to the limitation of the number of background galaxies .
we also use a statistical weight for each background galaxy in the context of @xmath43 where @xmath44 is the rms error for shear measurement of @xmath45th galaxy and @xmath46 is the softening constant variance . we set @xmath47 . the reduced shear at pixel position of @xmath48 is obtained by @xmath49 then , we invert the pixelized reduced shear ( equation [ eq : smshear ] ) with a weight of the inverse of the variance at each pixel to the lensing convergence field , based on the kaiser & squires inversion method @xcite .
we also map out distributions of optical luminosities and the number density of member galaxies with the same kernel with @xmath50 .
the luminosity and density maps are sensitive to luminous and high - density structures , respectively .
we selected red - sequence galaxies with @xmath51 as member galaxies .
the absolute magnitudes for member galaxies are calculated from the apparent magnitude using the @xmath52-correction for early - type galaxies .
we assume that all cluster member galaxies in the catalog are at a single redshift .
the resultant lensing @xmath53 field is shown in figures [ fig : opt+kappa ] and [ fig : kappa+den ] , with contours equi - spaced in units of @xmath54 reconstruction error , @xmath55 , above the @xmath54 level .
this mass distribution clearly exhibit a bimodal morphology in the central region ( @xmath56 ) , with two peaks reaching significance levels of @xmath57 and @xmath58 , respectively .
interestingly , these two peaks coincide with the two galaxy overdensities a1 and a2 hosting the two brightest cluster galaxies revealed in m08 , bcg1 , and bcg2 ( see also the top right and bottom left panels of figure [ fig : kappa+den ] ) .
we hereafter refer to these primary and secondary mass peaks as mc and mw , respectively .
the cluster mass distribution further exhibits some anisotropies at larger radii , mostly coinciding with peripheral optical clumps first revealed in m08 and presently confirmed ( see the bottom left panel of figure [ fig : kappa+den ] ) .
the most significant of these coinciding mass and optical peaks is @xmath59 at the optical clump , b. the mass distribution in the region of optical substructures c , d , and e , discovered by m08 , is similar to those of the number and luminous density , albeit low significance levels ( table [ tab : opticalclumps ] ) .
we measure luminosities within @xmath60 centering each @xmath53 peak ( table [ tab : opticalclumps ] ) , where a background region of @xmath61 centering bcg1 is used .
we find that luminosities and signal - to - noise ratios of @xmath53 are correlated .
we tried to measure model - independent mass for each optical subclump , following @xcite . however , since the result is sensitive to the choice of background region , we could not obtain reliable results .
an over - density region is further revealed from the density of galaxies whose colors are redder by @xmath62 at @xmath63 ( see the bottom right panel of figure [ fig : kappa+den ] ) .
the width of colors for these galaxies is @xmath64 .
distributed at the west of a2163 , these optical structures are associated with mass clumps of @xmath65 .
more generally speaking , the cluster mass distribution appears as spatially correlated with density distribution of its member galaxies , consistently with weak - lensing studies in other clusters @xcite .
we next compare the cluster mass contours with x - ray surface brightness and temperature maps derived from _ xmm - newton _ and _ chandra data _ analysis in b11 .
the left panel of figure [ fig : chan+kappa ] shows us a wavelet denoised map of the overall cluster atmosphere , with mass contours superimposed .
these two distributions clearly exhibit a spatial segregation , the x - ray emission appearing as unimodal and e - w elongated , with an emission peak located between the two mass peaks .
the right panel of figure [ fig : chan+kappa ] shows high - resolution details in a multi - resolution analysis of the _ chandra _ image . this analysis reveals a secondary cluster x - ray core , xw , separated from the main cluster emission peak , xc . as shown in b11
, this wedge - shaped feature ( red - curved line in the right panel of figure [ fig : chan+kappa ] ) appears as a stripped cool core embedded in the hotter icm of a2163-a and delimited in the westward direction by a cold front .
the evidence of this cold front let us infer the westward motion of this gas bullet along the elongated atmosphere of a2163-a .
interestingly , we now observe that the crossing core appears as preceded by the secondary mass peak , from which it might have been spatially separated during a subcluster accretion onto the main cluster .
this offset feature between the x - ray clump and mass peak is consistent with other cold front clusters on a merging phase @xcite .
it suggests a possibility of ram pressure stripping of the accreted cluster core ( see also section [ sec : dis ] ) .
it is further worth noticing that the main cluster mass peak , mc , also appears as slightly offset from the x - ray core , xc , so that the centroid of the gas core of xc and xw is at the intermediate position between the two mass peaks .
this ` double offset ' separating the gas and dark matter contents of both the infalling and main cluster cores presents some similarities to the fast accretion observable in 1e0657 - 56 @xcite , the so - called `` bullet - cluster '' .
to our knowledge , no other bimodal merger has been known so far to exhibit such a significant offset between the gas and dark matter components of its two components ( see , e.g. other examples in * ? ? ? * ) .
the northern subcluster a2163-b exhibits a spatial coincidence between x - ray , mass , and member galaxies , consistent with what is usually observed in pre - merging systems @xcite .
consistent with the lack of any interaction evidence found from icm thermodynamics between a2163-a and a2163-b , this spatial coincidence suggests that a2163-b is likely to be observed before interacting with a2163-a .
we conduct a tangential distortion study , which is a one - dimensional lensing analysis , in order to measure the total cluster mass .
the tangential distortion component of the reduced shear , @xmath66 , and the @xmath67 degree rotated component for individual galaxies ( @xmath45th galaxy ) are obtained by @xmath68 where @xmath69 is the position angle in the counter clockwise direction from the first coordinate axis on the sky .
then , the profiles of @xmath70 and @xmath71 are estimated with a statistical weight ( equation [ eq : ug ] ) , as follows : @xmath72 here , @xmath73 denotes the @xmath73th radial bin @xmath74 with a given bin .
the statistical error of @xmath75 in each radial bin is estimated as @xmath76 where the prefactor @xmath77 comes from the fact that @xmath44 in the rms is the sum of two distortion components .
we fit the tangential distortion profile with the universal profile proposed by navarro , et al .
( 1996 , hereafter nfw profile ) and a singular isothermal sphere ( sis ) halo model .
the nfw halo mass is a prediction of numerical simulations based cdm model .
the mass density profiles over a wide range of masses are well described in the form of @xmath78 where @xmath79 is the central density parameter and @xmath80 is the scale radius .
the asymptotic inner and outer slopes for nfw mass density are @xmath81 and @xmath82 , respectively .
the three - dimensional mass profile within a radius , @xmath83 , at which the mean density is @xmath84 times the critical mass density , @xmath85 , at the cluster redshift , is expressed by @xmath86 with @xmath87 the nfw halo mass is described by two parameters of the mass @xmath88 and the halo concentration @xmath89 .
the sis halo model is a solution of the collisionless boltzmann equation , and is specified by one parameter , the one - dimensional velocity dispersion @xmath90 , as follows , @xmath91 a three - dimensional mass for the sis model is given by @xmath92 the tangential distortion profile and the best - fit models are shown in figure [ fig : g+ ] .
we choose the central position determined by two - dimensional shear analysis which will be described in detail in section [ sec:2d ] .
we also measure the tangential shear profile with a center of bcg1 and fit them , but the results do not change significantly .
we can clearly find a curvature of the tangential shear profile .
the curvature makes it difficult to fit the profile with the sis model .
the @xmath93 for the sis model is @xmath94 ; therefore , the sis model can be strongly rejected ( @xmath95 level ) as a mass model .
on the other hand , the nfw mass model well expresses the curvature of the profile .
indeed , @xmath96 for the nfw model is @xmath97 .
the best - fit nfw parameters are found in table [ tab : mass ] .
the virial mass shows the massive cluster , @xmath98 , where the virial overdensity is @xmath99 .
our mass estimates are larger than those of a previous weak - lensing study ( table [ tab : mass ] ; * ? ? ?
* ) , using the shear catalog without excluding a contamination of member galaxies .
as shown in figures [ fig : opt+kappa ] and [ fig : kappa+den ] , the projected mass distribution of a2163 is complex and two mass peaks are significantly detected in the central region .
it is of prime importance for understanding cluster merger phenomena to measure masses of the main- and sub- clusters .
furthermore , since numerical simulations @xcite and observations @xcite have shown that a tangential shear profile is affected by significant substructures , taking into account substructures in modeling is also important for understanding such a lensing bias . in the mass measurement using the tangential shear profile ( section [ sec:1d ] ) , it is very difficult to distinguish which structure contributes in part to the tangential distortion signals , because the full lensing information from both the main and subclusters must be convolved to express the one dimensional distortion profile with respect to a given center @xcite .
the two dimensional shear pattern , on the other hand , enables us to easily model lensing signals by a superposition of lensing signals . in this section ,
we conduct two - dimensional shear fitting in order to measure the masses of three components ( the sub and main components for a2163-a and a2163-b ) revealed by m08 and b11 .
we pixelize the shear pattern into a regular grid of @xmath100 without any spatial smoothing procedure , whereas we adopted gaussian smoothing in the map making ( section [ sec : map ] ) .
the pixelized distortion signals and statistical weight , @xmath101 and @xmath102 , in the @xmath73th pixel are estimated with a weight function @xmath103 for each background source residing in the pixel ( see also equations .
( [ eq : g ] ) and ( [ eq : sig_g+ ] ) ) . the representative position for the @xmath73th pixel
is also estimated with a weight function @xmath103 .
the @xmath96 fitting is given by @xmath104 where @xmath105 is the parameters and @xmath106 is the error covariance matrix of shape measurements in the form of @xmath107 . here
, @xmath108 is a kronecker delta function and @xmath109 is the statistical error of the pixelized shear @xcite . we first consider a single mass model of the nfw profile in order to compare the mass estimates by tangential shear measurement .
we here treat the center of nfw mass ( @xmath110 , @xmath111 ) as a parameter . in total
, we use four parameters ( @xmath112 , @xmath113 , @xmath110 , and @xmath111 ) for fitting .
we adopt the markov chain monte carlo method with standard metropolis - hastings sampling .
we restrict the sampling range of @xmath114 , @xmath115 .
we refer to the mean of the posterior probability distribution of each parameter .
the resultant masses are consistent with the tangential shear measurements ( table [ tab : mass ] ) .
the central position of mass is consistent with the peak position of the mc clump in the mass map .
the mc clump , which is associated with bcg1 , is therefore likely to be the main component .
we next add a mass model for the mass clump mw to the main cluster . from x - ray and optical spectroscopic studies
( m08 and b11 ) , the mass clump mw is likely to be a merging substructure in a2163-a . since cluster substructure size is not determined by the virial theorem but by the strong tidal force of the main cluster ( e.g. , * ? ? ?
* ) , we adopt the truncated sis ( tsis ) model @xcite to describe the mw clump . the tsis model is an extreme case of the truncation ; mass density becomes zero at a radius @xmath116 .
the tsis mass profile is expressed as @xmath117 the subclump mass for the tsis model is estimated as @xmath118 the tsis model is specified by one - dimensional velocity dispersion @xmath119 and the truncation radius @xmath116 .
we have an additional four parameters ( @xmath119 , @xmath116 and centers ) for the tsis model and a total of eight parameters for the fitting .
we assume that the redshift of the mw clump is the same as that of the main cluster .
we adopt @xmath120 and @xmath121 , where @xmath122 and @xmath123 are peak coordinates appearing in a weak - lensing mass map .
the resultant mass and central positions are shown in table [ tab : mass2 ] .
the centroid of the mw clump is consistent with the peak found in the mass map .
the truncation radius , @xmath124 , is an intermediate size compared to the virial radius of the main cluster @xmath125 . since the sub - cluster size after some impacts
is significantly decreased by the strong tidal field , such a large truncation radius might suggest that the mw clump is a merging sub - cluster at the first impact .
next , we take into account the northern component a2163-b ( m08 and b11 ) , which contains luminous galaxies ( optical clump b ) and an x - ray emitting core .
we consider two possibilities for the dynamic state of a2163-b ; one scenario is the pre - merger phase that a2163-b is infalling toward a2163-a , and the other is that a2163-b has already undergone a merging event with the main cluster of a2163-a .
b11 found no evidence of the close interaction between a2163-a and a2163-b , and therefore concluded that they are likely to be separated more than @xmath126 aligning along the line of sight .
however , in m08 , a filament of faint galaxies was detected along a north / south axis between a2163-a and a2163-b which might suggest a previous interaction between the two components .
although this post - merger hypothesis is very unlikely from the x - ray approach , the origin of this faint galaxy filamentary structure is still an open issue . in this paper , we investigate two possibilities in the fitting . if a2163-b is physically separated from a2163-a , its mass profile is not likely to be affected by the strong tidal field of a2163-a and we therefore adopt the nfw model for the first scenario . here ,
the halo concentration is ill constrained because the shear signals from a2163-b are smaller than those from a2163-a , and it is difficult in the environment of the massive cluster to find the curvature of the distortion profile , as shown in figure [ fig : g+ ] .
we therefore assume the mass - concentration relation of @xmath127 @xcite . here
, we assume a redshift of a2163-b , @xmath128 , is the same as that of a2163-a .
this assumption is justified by both the dynamical analysis ( m08 ) and the photometric redshift estimates by @xcite .
we also parameterize the center of a2163-b , using 13 parameters in total .
the resultant virial mass is @xmath129 ( table [ tab : mass2 ] ) .
the mass @xmath130 within a radius @xmath126 , at which the mean density is @xmath131 times the critical density , agrees well with x - ray estimates @xmath132 ( b11 ) .
we changed the normalization of the mass -concentration relation by @xmath133 and conducted fittings , but the best fit of the virial mass for a2163-b changes only by @xmath134 and @xmath135 .
they all are consistent within errors .
we also found that the mc and mw masses are almost unchanged after adding the mass model of a2163-b .
the resultant mass ratios of mc ( main cluster of a2163-a ) , mw ( sub cluster of a2163-a ) and a2163-b are found to be @xmath136 .
we consider the second scenario that a2163-b is an ongoing merger with a2163-a .
we here use the tsis model for a2163-b , following the case of the substructure mw .
the truncation radius @xmath137 is obtained , while the concentration of the nfw mass model is not well constrained .
this is because the tangential distortion profile for the truncation model outside the truncation radius is proportional to the inverse square of the radius ( @xmath138 ) , which is different from that of the nfw mass model for the main cluster .
we obtain the substructure mas @xmath139 .
the mass discrepancy between the two models is small : @xmath140 and @xmath141 with uncertainties of nfw and tsis masses , respectively .
the two mass models ( nfw and tsis ) obtained solely by shear data are , therefore , acceptable for a2163-b .
the total mass in each model is consistent within errors with tangential shear measurement , which indicates that a substructure effect is likely to be less significant .
due to its exceptionally high mass and dynamical activity , a2163 has long been an interesting test case for cluster mass measurements .
we here compare the weak lensing cluster mass , @xmath142 ( see section [ sec:1d ] and figure [ fig : mhe_vs_mwl ] ) , with dynamical and x - ray mass estimates provided by m08 and b11 .
as previously mentioned , weak lensing mass measurements do not require any assumption of cluster dynamical state .
they consequently provide us a unique opportunity to investigate effects of the virial theorem hypothesis for member galaxies or the gas h.e .
assumption on mass measurement precision .
we first compare weak - lensing mass with a dynamical one .
m08 estimates dynamical total mass from the velocity dispersion , @xmath143 , under assumptions of the virial theorem , spherical symmetry , and no internal structure . the virial radius in dynamical mass estimation is defined as the three - dimensional radius of harmonic radius , corresponding to a coherent length of galaxy separations , which is different from that of the weak - lensing mass measurement .
we here therefore compare @xmath144 enclosed in the radius @xmath145 within which the mean density is @xmath146 times the critical density @xmath147 .
dynamical mass , @xmath148 , extrapolated with a hernquist profile @xcite , is higher than the weak - lensing mass measured by both one- and two- dimensional analyses ( table [ tab : mass ] ) .
the discrepancy might be due to a difference of models .
we would require a more detailed dynamical model considering uncertainty of the anisotropy in the velocity distribution @xcite . on the other hand , we calculate one - dimensional velocity dispersion corresponding to the virial mass for the nfw model and obtain @xmath149 assuming an isotropic velocity dispersion , which agrees well with the velocity dispersion along the line - of - sight ( m08 ) .
they are higher than the one - dimensional velocity dispersion , @xmath150 , of the sis model which is ruled out at the @xmath95 level by one - dimensional weak - lensing analysis .
we note that this is a rough comparison , because the observed line - of - sight velocity dispersion is calculated with a weight of mass density @xcite .
we next compare x - ray masses using h.e . assumption .
the impact of the ongoing cluster merger event in the central region of a2163 on hydrostatic equilibrium ( h.e . ) of the hot gas has been investigated in b11 . to do so
, an average cluster mass profile has been extracted from density and temperature profiles of the overall cluster atmosphere , in addition to two supplementary profiles extracted assuming h.e . in the complementary regions located behind and ahead the gas core currently crossing a2163 ( see also sect . [
sec : mass ] and figure 7 of b11 ) .
these profiles have been compared with our nfw lensing profile in figure [ fig : mhe_vs_mwl ] .
we first find that h.e .
mass profiles extracted in the eastern cluster side and overall cluster strongly exceed the nfw profile at an overdensity radius of @xmath126 , and beyond .
interestingly , we find instead that the h.e .
mass profile extracted in the western cluster sector fully agrees with the weak- lensing profile in the radii range @xmath151@xmath126 .
mass estimates corresponding to this analysis at the overdensity @xmath152 are reported in table [ tab : mass ] .
adopting overdensity radii determined by each measurement method , we observe that the h.e .
mass in all and eastern sectors are @xmath153 ( @xmath154 level ) and @xmath155 times ( @xmath156 level ) higher than the weak lensing mass , respectively , while the one in the western sector is in a good agreement with weak lensing one . from this comparison
, we may infer that h.e .
assumption is only realistic in the western side of the cluster outer radii .
the eastern cluster side is instead likely to have been shock heated by the ongoing subcluster accretion , yielding a strong departure of this cluster region and thus of the overall cluster atmosphere from h.e .. one concern in cluster cosmology is the search for x - ray mass proxies relying on well - calibrated scaling relations coupling the cluster gas properties with total masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite have proposed a new mass proxy , the so - called , quasi - integrated gas pressure , @xmath158 .
this quantity has been suggested by numerical simulations to exhibit low intrinsic scatter regardless of the cluster dynamical state , in particular since deviations of temperature and gas mass from the normalization of the mass scaling relations partly anti - correlate during cluster mergers .
the cluster mass , @xmath159 , has been estimated using the @xmath157 proxy in b11 , assuming the @xmath159@xmath157 scaling relation to be calibrated from hydrostatic mass estimates in a nearby cluster sample observed with _
xmm - newton _ @xcite .
this mass estimate , @xmath160 , yields a 20 @xmath161 excess with respect to our weak - lensing mass . while marginally significant ( 1.1@xmath162 significance level ) , this mass discrepancy slightly exceeds the @xmath163 scatter of the @xmath157 proxy suggested by numerical simulations in @xcite . @xcite have investigated a dynamical dependence on mass observable scaling relations , based on _ xmm - newton _ x - ray observables @xcite and weak- lensing masses @xcite .
this study found the normalization of the @xmath159@xmath157 scaling relation to be lower for disturbed clusters than for undisturbed clusters , contradicting the prediction of normalization independence in cluster dynamical states .
mass estimates of a2163 relying on the undisturbed and disturbed cluster scaling relation calibrated by @xcite yield @xmath164 and @xmath165 , respectively .
interestingly , the mass estimate with the scaling relation for disturbed clusters fully agrees with weak - lensing mass , while the undisturbed cluster estimate yields a higher value , consistent with the @xmath157 estimate of b11 .
since a2163 is an ongoing merger , the normalization for disturbed clusters gives a better result . in order to constrain the cosmological parameters ( e.g ,
* ; * ? ? ?
* ) , it is of prime importance to construct and calibrate a low - scatter mass proxy .
recently , millennium gas simulations @xcite predicted that the temperature and gas - mass deviations are positively correlated , which contradicts @xcite .
it is also important to construct a low - scatter mass proxy using solely observational data , based on intrinsic covariance measurement and principal component analysis @xcite .
further systematic studies , including both numerical simulation and observational study , are required for this .
from the inversion of a background galaxy shear pattern , we revealed a bimodal mass distribution in the central region of a2163 and various anisotropies at larger radii , the most significant of which corresponding to the northern subcluster a2163-b .
modeling the underlying three dimensional distribution of the cluster mass with three components further allowed us to constrain the mass ratio between the two central components to 1:8 , and attribute comparable mass values to the central and northern subclusters .
the spatial coincidence between x - ray peak , mass and member galaxy distribution in a2163-b suggests that a2163-b did not yet interact with a2163-a .
while coinciding with the bimodal galaxy distribution revealed in m08 , the central mass distribution appears instead as spatially segregated from the brightest x - ray emitting gas region . in particular , the gas bullet suggested to cross the cluster atmosphere from the evidence of a cold front , in b11 , appears as preceded by the secondary mass peak to a projected distance of @xmath166 along its westward direction of motion . assuming an infalling subcluster has experienced a transient separation of its gas and dark matter components as a result of ram pressure stripping
, this configuration would require the gas core to have survived tidal distortions exerted by the main cluster , and the core temperature to be consistent with the subcluster mass . in the following ,
we discuss how these conditions allow us to put some additional constraints on the kinematics of the ongoing subcluster accretion .
one possibility for explaining the offset between mass and gas distributions is the ram - pressure stripping : if the ram - pressure force on the x - ray core is stronger than the gravity around the core region , the gas is stripped from its gravitational potential . this condition is expressed as @xmath167 @xcite . here , @xmath168 is the radius of the x - ray core , @xmath169 is the spherical mass of the subcluster within the radius @xmath170 , @xmath171 are the density of the core and its surrounding gas , respectively , and @xmath172 is the velocity of the core in the center - of - mass frame .
@xmath173 is a fudge factor at an order of unity .
this is why the ram - pressure stripping is not effective due to a kelvin - helmholtz instability , magnetic fields , and shock heating .
b11 shows a lower limit of the density jump , @xmath174 , at the west edge of the core ( red - curved line in figure [ fig : chan+kappa ] ) .
we consider the ram - pressure stripping condition on the sub cluster .
the substructure mass inside @xmath168 is calculated from the tsis model obtained by the weak - lensing analysis . since @xmath175 ,
the condition ( [ eq : pram ] ) is free from the core size .
although we consider the internal structure of gas , the resultant condition does not significantly change .
figure [ fig : pram ] plots the ram - pressure stripping condition @xmath176 as a function of infall velocity @xmath172 , where @xmath177 gives the ratio of the right - hand side to the left - hand side of equation ( [ eq : pram ] ) .
we adopt @xmath178 .
m08 found the gradient of line - of - sight velocity @xmath179 , giving the lower - limit of infall velocity .
the infall velocity is expected from the masses to have an order of @xmath180^{1/2}\sim 2100 ~{\rm km~s^{-1 } } \label{eq : vmerger}\end{aligned}\ ] ] @xcite , where @xmath112 and @xmath181 are the virial mass and radius , respectively .
we here used observed truncated mass and radius for the sub cluster .
the case of @xmath182 satisfies the ram - pressure stripping condition for all velocities above @xmath183 .
the case of @xmath184 also satisfies the condition around @xmath185 .
therefore , the x - ray core initially associated with the sub cluster is easily stripped away from its central region . however , we must keep in mind that the estimated density jump might give the lower limit , due to the deprojection method we applied . in this case , a requirement of the ram pressure condition gives @xmath186 .
we next discuss the destruction process of the gas core by the strong tidal field of the main cluster .
numerical simulation @xcite has shown that the lifetime for a gas satellite is shorter than that for a dark matter one . in particular , the lifetime for a gas core decoupled from dark matter potential is much shorter than that of a dark matter satellite .
we roughly estimate the gas core disruption by the tidal force of the main cluster .
as shown above , the gas core could easily be stripped away from the subcluster s core region .
we therefore consider the stripped core which is composed of the gas , to be free from the gravity of the subcluster . the gas model for the core
is adopted as a single power - law density profile as an approximation , @xmath187 where the normalization @xmath188 and the slope @xmath46 are the model parameters .
we use the position of cold front @xmath189 ( red - curved line of the right panel of figure [ fig : chan+kappa ] ) .
the tidal radius at which the core density is truncated is estimated by the force equivalence between the internal gravity and external tides @xcite , @xmath190^{1/3 } , \label{eq : tidal}\end{aligned}\ ] ] in the limit of @xmath191 and @xmath192 . here
@xmath116 is the tidal radius and @xmath193 is the core position from the center of the main cluster ; that is , an impact parameter .
@xmath194 is a fudge factor because hydrodynamic instabilities effectively destruct the gas core .
we assume @xmath195 in order to consider the tidal disruption only .
we also assume that the density profile within a tidal radius dose not modify before and after the tidal stripping .
the main cluster mass profile uses the best - fit nfw model in the case of nfw+tsis+nfw ( table [ tab : mass2 ] ) .
we first calculate the minimum tidal radius @xmath196 as a function of the density normalization and its slope and impact parameter .
the left panel of figure [ fig : taidal ] shows a distribution of the minimum tidal radius of the gas core in the parameter plane of the density normalization and the slope .
the contours denote the minimum tidal radius in the range of @xmath197 .
the point denotes the normalization and slope parameters for the core xw at the cold front . in the parameter plane ,
the minimum tidal radius is much smaller than the observed radius @xmath198 , indicating that we can not observe a remnant of the gas core if a2163-a is a system of close encounter .
we next try to constrain the impact parameter , @xmath193 , so as to realize the observed radial size of the core xw ( @xmath199 ) through equation .
( [ eq : tidal ] ) .
the right panel shows a distribution of the impact parameter in the parameter plane of the density normalization and the slope .
the contours present the impact parameters @xmath200 to explain the core size .
the observed x - ray core requires the impact parameter @xmath201 .
if a subcluster collides into the main cluster with this impact parameter , the central region of the main cluster is significantly affected by cluster merger .
it does not conflict with the disturbed core structures and the complex temperature distribution ( figure [ fig : chan+kappa ] ) .
since this estimation is based on the gravitational process only , we need to keep in mind a possibility that hydrodynamic instabilities shorten the lifetime of gas .
we emphasize that this method demonstrates only one of the methods for joint x - ray and weak - lensing analysis . assuming the cool gas core xw and the secondary mass peak mw represent the separated components of a formerly accreted subcluster
, the intrinsic temperature of xw should agree with the expectation of a self - similar subcluster temperature corresponding to the mass of mw .
given our multi - component mass distribution ( nfw+tsis+nfw ; see table 2 ) , we discuss the consistency of this hypothesis with projected icm temperatures derived near the cool gas core xw in b11 .
the cool gas core xw being located a relatively close projected distance from the main cluster emission peak xc , average icm temperatures measured along a line of sight intercepting xw would consist of a linear combination of the main and subcluster temperatures .
these temperatures can be predicted within the radial range [ 0.2 - 0.5]@xmath202 , from the @xmath203~r_{500}}$ ] scaling relation calibrated by @xcite , in a sample of 12 local clusters analyzed in both x - ray and weak - lensing . to perform this estimation , we first measure the main cluster mass , @xmath204 , by multi - component analysis ( nfw+tsis+nfw ) , within the overdensity radius @xmath126 .
we further determine the subcluster mass , @xmath205 , from its current mass , @xmath206 , assuming the subcluster mass profile to have followed a universal nfw radial distribution before the ongoing accretion , and the profile concentration to be related to @xmath207 by the halo mass - concentration dependence of @xcite .
these assumptions would yield a cluster and subcluster temperature prior to the accretion of @xmath208 and @xmath209 , respectively , where the first , second and third errors are from mass measurement error , normalization error of scaling relation , and intrinsic scatter in the relation .
assuming the main and subcluster emissivities to be comparable from a rough analysis of the x - ray image and weighting these temperatures following a scheme proposed in @xcite , an average icm temperature measured in the direction of the cool core would reach a value of @xmath210 . despite being consistent with uncertainties in our mass estimates ,
this value is considerably lower than projected temperatures measured in the cool gas core xw in the _ xmm - newton _ temperature map and the _ chandra _ temperature profile of b11 ( @xmath211 ) . among the possible origin for this discrepancy
, the cool gas core might have been heated from its virial temperature while crossing the main cluster atmosphere , possibly due to mixing with the main cluster atmosphere or a reverse shock , presumably at an earlier stage of its accretion and prior to the formation of the cold front .
[ [ section ] ] in summary , the density of the main cluster atmosphere and the velocity of its presumably free - falling subcluster are large enough to have separated the subcluster gas and dark matter components through ram - pressure stripping .
moreover , the survival of the subcluster gas core against tidal forces exerted by the main cluster implies the subcluster accretion to have occurred with a non - zero impact parameter . comparing the average x - ray temperature in the direction of the cool core to its expectation from projection of the cluster and subcluster virial temperatures yields however a mild inconsistency , suggesting the cool core was partially heated while crossing the main cluster atmosphere .
deeper x - ray observations should help better constrain the shape , temperature and mass of the cool core and refine this accretion scenario .
we presented the weak - lensing analysis of the merging cluster a2163 using the subaru / suprime - cam and cfht / mega - cam data , measured cluster mass by one - dimensional tangential shear analysis , and measured three components masses by multi - component analysis of the two - dimensional shear pattern .
based on complementary x - ray , dynamical and weak - lensing datasets , we also discussed the centroid offset between weak - lensing mass and gas core .
our main results are summarized below . *
the projected mass distribution shows a bimodal structure in the central part of a2163-a .
the overall mass distribution appears to be similar to the member galaxy one , whereas both mass and member galaxy distributions are completely different from the icm one .
this is consistent with a previous study of seven merging clusters at various dynamical state @xcite .
in particular , we found a clear offset between the gas core associated with the cold front and sub - cluster mass peak .
the offset was reported in all cold - front clusters previously conducted by weak - lensing analysis @xcite .
the gas core is also offset from the main cluster mass peak , like the bullet cluster @xcite .
* a two dimensional shear analysis has enabled us to measure the mass of each of the three major components in a2163 , including the two components of a2163-a , and the northern subcluster a2163-b .
the central subcluster in a2163-a is well described by the tsis model , giving the mass @xmath212 .
* the mass of a2163-b is comparable to the sub cluster of a2163-a , which is in good agreement with the x - ray estimate of b11 , assuming h.e .
the mass , member galaxies , and gas distributions are similar to one another , as reported in pre - merging clusters @xcite , suggesting that a2163-b did not yet interacted with a2163-a . * the gas bullet suggested to cross the cluster atmosphere from the evidence of a cold front , in b11 , appears to be preceded by the secondary mass peak to a projected distance of @xmath166 .
we show that the density of the main cluster atmosphere and the free fall velocity of an incoming subcluster with which mass is about one - eighth of main cluster virial mass are large enough to have separated the dark matter and gas component of this subcluster through ram - pressure stripping . following this scenario ,
the survival of the cool core against tidal forces exerted by the main cluster lets us infer that the subcluster must have been accreted with a non - zero impact parameter , reaching typical values of @xmath213 . assuming the cool gas core and the secondary mass peak represent the separated components of the formerly accreted subcluster , the projected core temperature appears higher than expected from our estimates of the main and subcluster virial temperatures .
subsequent to its separation with dark matter , the crossing cool core may thus have been shock heated or disturbed by hydrodynamical instabilities yielding a partial mixing with the main cluster atmosphere . *
dominated by the most massive component of a2163-a , the overall mass distribution in a2163 is well described by a universal nfw profile as shown by a tangential distortion analysis , while the sis profile is strongly rejected ( @xmath95 confidence level ) .
the virial mass for the nfw mass model is @xmath214 .
* we compare the weak - lensing nfw mass with dynamical and x - ray h.e .
ones . the weak - lensing mass is lower than the dynamical one with assumptions of the virial theorem and spherical symmetry mass distribution .
this might be due to the dynamical mass model , because the line - of - sight velocity dispersion expected from the nfw mass agrees well with the spectroscopic result ( m08 ) .
mass , @xmath159 , in the western sector showing no hot gas , is in good agreement with the weak - lensing one , whereas the one in the eastern sector showing hot gas is higher at the @xmath95 level .
this indicates that the merger shock heating leads to an overestimation of h.e .
the mass proxy , @xmath157 , through the mass observable scaling relation for disturbed clusters @xcite gives a mass estimate similar to the weak - lensing mass .
no acknowledges yoichi oyama , keiichi umetsu , motoki kino , keiichi asada , makoto inoue and alberto cappi for helpful discussions . we thank the anonymous referee for the useful comments that led to improvement of the manuscript .
we also thank g. soucail for giving us cfht @xmath13-band data .
we are also grateful to n. kaiser for developing the publicly available imcat package . h.b and p.m acknowledge support by nasa grants nnx09ap45 g and nnx09ap36 g grant asi - inaf i/088/06/0 and asi - inaf i/009/10/0 .
arnaud , m. et al.,1992 , apj , 390 , 345 .
arnaud , m. , pratt , g. w. , piffaretti , r. , bhringer , h. , croston , j. , h. , & pointecouteau , e. , 2010 , a&a,517,a92 .
bardeau , s. , soucail , g. , kneib , j .-
p . , czoske , o. , ebeling , h. , hudelot , p. , smail , i. , & smith , g. p.,2007 , a&a , 470 , 449 .
bartelmann , m. , schneider , p. , 2001
, phys . rep .
, 340 , 219 .
becker , m. r. & kravtsov , a. v. , 2010 , apj submitted ( arxiv:1011.1681 ) .
binney , j. , & mamon , g. , a.,1982,mnras,200,361 .
bourdin , h. , arnaud , m. , mazzotta , p. , pratt , g. w. , sauvageot , j .- l . , martino , r. , maurogordato , s. , cappi , a. , ferrari , c. , benoist , c. , 2011 , a&a 527 , a21 .
( b11 ) broadhurst , t. , takada , m. , umetsu , k. , et al .
, 2005 , apjl , 619 , 143 . k. & setti , g. , 2009,a&a,507,661 .
clowe , d. , et al .
2006 , a&a , 451 , 395 cypriano , e. s. , sodr , l. , jr . , kneib , j .-
p . , & campusano , l. e. 2004 , , 613 , 95 duffy , a. , r. , schaye , j. , kay , s. , t. & dalla v. , c. , 2008,mnras,390,l64 .
erben , t. , van waerbeke , l. , bertin , e. , mellier , y. , & schneider , p. 2001
, a&a , 366 , 717 .
elbaz , d. , arnaud , m. , boehringer , h.,1995 , a&a,293 , 337 .
feretti , l. , fusco - femiano , r. , giovannini , g. & govoni , f.,2001,a&a , 373 , 106 .
feretti , l. , orr , e. , brunetti , g. , giovannini , g. , kassim , n. & setti , g. 2004 , a&a,423,111 .
gray , m. e. , taylor , a. n. , meisenheimer , k. , dye , s. , wolf , c. & thommes , e.,2002 , apj,568,141 .
gavazzi , r. , fort , b. , mellier , y. , pell , r. , dantel - fort , m. , 2003 , a&a , 403 , 11 hernquist , l.,1990 , apj,356 , 359 .
hoekstra , h. , 2007,mnras,379 , 317 .
ilbert , o. , et al .
2009 , apj , 690 , 1236 .
kaiser , n. & squires , g. 1993 , apj , 404 , 441 .
kaiser , n. , squires , g. , broadhurst , t. , 1995 , apj , 449 , 460 .
kawaharada , m. , okabe , n . , umetsu , k. , takizawa , m. , matsushita , k. , fukazawa , y. hamana , t. , miyazaki , s. , nakazawa , k. , & ohashi , t. , 2010,apj,714 , 423 .
kravtsov , a. , v. , vikhlinin , a. , a. , & nagai , d. , 2006,apj,650,128 .
la barbera , f. , merluzzi , p. , busarello , g. , massarotti , m. & mercurio , a. 2004 , a&a,425 , 797 .
okas , e. l. & mamon , g. , a. , 2001 , mnras,321,155 .
mahdavi , a. , hoekstra , h. , babul , a. , balam , d. , d. , & capak , p. , l. , 2007,apj,668,806 .
mahdavi , a. , hoekstra , h. ; babul , a. , henry , j. p. , 2008
, mnras,384,1567 markevitch , m. & vikhlinin , a. , 2001,apj,563,95 .
markevitch , m. & vikhlinin , a. , 2007 , phr,443,1 .
maurogordato , s. , cappi , a. , ferrari , c. , benoist , c. , mars , g. , soucail , g. , arnaud , m. , pratt , g. w. , bourdin , h. & sauvageot , j .- l . , 2008 , a&a , 481 , 593 ( m08 ) .
mazzotta , p. , rasia , e. , moscardini , l. , & tormen , g. 2004 , , 354 , 10 .
meneghetti , m. , rasia , e. , merten , j. , bellagamba , f. , ettori , s. , mazzotta , p. , dolag , k. & marri , s. , 2010 , a&a , 514 , a93 .
merten , j. et al .
, 2011 , mnras accepted ( arxiv:1103.2772 ) navarro , j. f.,frenk , c. s. & white , s. d. m. 1996 , apj , 462 , 563 .
nord , m. , et al .
2009 , a&a , 506 , 623 planck collaboration : ade , p. , a. , r. , , aghanim , n. , arnaud , m. et al . 2011 , arxiv:1101.2026 oguri , m. , takada , m. , okabe , n. , & smith , g. , p. , 2010
, mnras , 405,2215 .
okabe , n. & umetsu , k. 2008 , pasj , 60 , 345 okabe , n. , okura , y. , & futamase , t. , 2010 , apj , 713 , 291 .
okabe , n. , takada , m. , umetsu , k. , futamase , t. & smith , g. , p. , 2010
, pasj , 62 , 811 .
okabe , n. , zhang , y .,-
y . , finoguenov , a. , ; takada , m. , smith , g. , p. , umetsu , k. , & futamase , t. , 2010 , apj , 721 , 875 .
ouchi , m. , et al .
2004 , apj , 611 , 660 .
radovich , m. , puddu , e. , romano , a. , grado , a. & getman , f. , 2008 , a&a,487 , 55 .
ricker , p. m. & sarazin , c. l. 2001 , apj , 561 , 621 .
schneider , p. , `` _ weak gravitational lensing _ '' , lecture notes of the 33rd saas - fee advanced course , g. meylan , p. jetzer & p. north ( eds . ) , springer - verlag : berlin , p.273 shan , h. , qin , b. , fort , b. , tao , c. , wu , x .-
p . , & zhao , h. 2010 , , 406 , 1134 stanek , r. , rasia , e. , evrard , a. , e. , pearce , f. , & gazzola , l. , 2010 , apj,715,1508 .
tormen , g. , diaferio , a. , & syer , d. , 1998,mnras , 299 , 728 .
tormen , g. , moscardini , l. & yoshida , n. , 2004 , mnras , 350 , 1397 .
takizawa , m. , 2006 , pasj , 58 , 925 .
umetsu , k. & broadhurst , t. , 2008 , apj , 684 , 177 .
umetsu , k. et al .
2009 , apj , 694 , 1643 .
umetsu , k. et al . 2010 .
apj , 714 , 1470 .
vikhlinin , a. , burenin , r. a. , ebeling , h. , forman , w. r. , hornstrup , a. , jones , c. , kravtsov , a. v. , murray , s. s. , nagai , d. , quintana , h. , & voevodkin , a. 2009a , apj , 692 , 1033 .
vikhlinin , a. , kravtsov , a. v. , burenin , r. a. , ebeling , h. , forman , w. r. , hornstrup , a. , jones , c. , murray , s. s. , nagai , d. , quintana , h. , voevodkin , a. , 2009b , apj,692,1060 .
watanabe , e.,takizawa , m. , nakazawa , k. , okabe , n. , kawaharada , m. , babul , a. , finoguenov , a . , smith , g. , p. , & taylor , j. , 2011 , pasj , 63 , 357 .
zhang , y .- y . , finoguenov , a. , bhringer , et al .
2008 , a&a , 482 , 451 .
zhang , y .- y . , okabe , n. , finoguenov , a. , et al .
2010 , apj , 711 , 1033 | we present a weak - lensing analysis of the merging cluster a2163 using subaru / suprime - cam and cfht / mega - cam data and discuss the dynamics of this cluster merger , based on complementary weak - lensing , x - ray , and optical spectroscopic data sets . from two - dimensional multi - component weak - lensing analysis , we reveal that the cluster mass distribution is well described by three main components including a two component main cluster a2163-a with mass ratio 1:8 , and its cluster satellite a2163-b .
the bimodal mass distribution in a2163-a is similar to the galaxy density distribution , but appears as spatially segregated from the brightest x - ray emitting gas region .
we discuss the possible origins of this gas - dark matter offset , and suggest the gas core of the a2163-a subcluster has been stripped away by ram pressure from its dark matter component .
the survival of this gas core to the tidal forces exerted by the main cluster lets us infer a subcluster accretion with a non - zero impact parameter .
dominated by the most massive component of a2163-a , the mass distribution of a2163 is well described by a universal navarro - frenk - white profile as shown by a one - dimensional tangential shear analysis , while the singular - isothermal sphere profile is strongly ruled out . comparing this cluster mass profile with profiles derived assuming intracluster medium hydrostatic equilibrium ( h.e . ) in two opposite regions of the cluster atmosphere has allowed us to confirm the prediction of a departure from h.e . in the eastern cluster side , presumably due to shock heating .
yielding a cluster mass estimate of @xmath0 , our mass profile confirms the exceptionally high mass of a2163 , consistent with previous analyses relying on the cluster dynamical analysis and @xmath1 mass proxy . |
in the last few years , great strides have been made towards understanding the nature of the faint sub - mjy and microjy radio source population .
most of these studies have employed connected arrays , such as the vla _
richards et al .
1998 , richards 2000 ; merlin _
e.g. _ muxlow et al .
1999 ; atca _
e.g. _ norris et al .
2001 and the wsrt _
garrett et al . 2001 ) .
these observations typically integrate for many days on relatively small patches of sky , targeting well known `` deep field '' regions , rich in complimentary observations at sub - mm , mid - ir , near - ir , optical and x - ray wavelengths .
such radio observations reach r.m.s .
noise levels @xmath0 few microjy / beam , encountering source densities of @xmath1 per square arcmin at 1.4 ghz . until recently the idea that vlbi observations could detect even one solitary , compact radio source in these narrow , radio - quiet , deep field regions would have been tantamount to a declaration of professional suicide . however , such observations are now possible due to : ( i ) increasing individual telescope sensitivity coupled with sustained , high data rate recording systems , ( ii ) the simultaneous employment of both phase - referencing and wide - field imaging techniques , and ( iii ) the availability of more capable correlators , matched by equally impressive desktop data analysis facilities . in this paper
i will review the current status of deep , wide - field , vlbi continuum imaging - focusing on the technique and the science it enables .
i will also look forward to future developments ( e.g. @xmath2evn ) these promise to completely transform our current capabilities , in a way that will open - up entirely new areas of vlbi research .
but first i begin with a summary of the properties of the faint , sub - mjy and microjy radio sky .
fig . 1 presents a deep 1.4 ghz wsrt image ( garrett et al .
2000a ) of the hubble deep field ( hdf williams et al .
one of the simplest but most striking observations is the close correspondence between faint wsrt 1.4 ghz detections and iso @xmath3 m detections . in particular , the vast majority of these distant sources that are common to both the wsrt and iso appear to follow the well - known fir - radio correlation ( garrett 2002 ) .
this fir - radio correlation was first observed with respect to `` normal '' star - forming galaxies in the local universe ( e.g. condon 1992 ) .
its extension to these much more luminous and more distant systems in the hdf ( the wsrt / iso sample has a median redshift of 0.7 ) is convincing evidence that the majority of the faint radio source population are identified as star forming galaxies with star formation rates ( sfr ) often well in excess of 100 m@xmath4/yr .
this observation ( see fig .
[ fir - rc ] ) is backed up ( richards et al .
1998 , muxlow et al . 1999 ) by the measured angular size of the radio sources , their radio morphology / steep radio spectra , and their optical morphology ( the latter often showing evidence for galaxy interactions ) .
a picture is thus developing of a radio sky that literally `` lights up '' at microjy flux density levels , with the bulk of the radio emission being directly related to massive star - formation i.e. supernovae events and the global acceleration of cosmic ray electrons via propagating shocks that are associated with these events .
note , however , that these conclusions are dominated by the faintest ( and most numerous ) microjy radio sources , in particular the agn fraction is observed to increase rapidly at higher ( sub - mjy ) flux density limits . in addition , there is a significant fraction of these faint radio sources for which little is known , these are the optically faint radio source population believed to be high redshift galaxies in which the dust content is large enough to completely obscure the ir , optical , uv and even soft x - ray emission .
there seems to be a correspondence between this population and the faint scuba source population ( e.g. hughes et al . 1998 ) .
the latter are now considered to be responsible for the vast bulk of the star formation in the early universe but their detailed properties remain largely unknown . in particular , it is unclear whether the radio and fir emission arise via agn , star formation activity or some combination of both phenomena .
in principle extremely sensitive global vlbi observations have sufficient resolution to distinguish between agn and starburst activity in these optically faint radio sources .
a taste of what may be possible is provided by the recent evn observations of the hubble deep field .
on 12 - 14 november 1999 the evn conducted the first pilot vlbi `` blank field '' observations of the radio sky .
the field chosen was the hdf - n an area that is just about as empty and undistinguished as the radio sky gets .
the brightest source in the @xmath5 arcminute radial field of view was an fr - i radio galaxy with a total wsrt 1.4 ghz flux density of @xmath6 mjy .
the data were recorded at a rate of 256 mbits / sec for 32 hours a sustained capability that is unique to the evn ( and has recently been extended to 512 mbits / sec ) . observing in external phase - reference mode ,
a total of @xmath7 hours of `` on - source '' data were collected . with a resolving beam area 1 million times smaller than the wsrt hdf - n observations , the evn surveyed an area of about 12 square arcminutes .
six hdf - n radio sources were thus targeted simultaneously ( using wide - field imaging techniques
see garrett et al .
the final naturally weighted images have an r.m.s
. noise level of @xmath8jy / beam much larger than that expected from thermal noise considerations ( @xmath9jy / beam ) .
the images are probably limited by the inclusion of poorly calibrated or completely corrupt data - difficult to identify in this case , and not uncommon with external phase - calibration during observations made at solar maximum .
nevertheless , the evn simultaneously detected three radio sources above the @xmath10jy ( @xmath11 ) limit , in the inner part of hdf - n region ( see fig [ evn_im ] ) .
these include : vla j123644 + 621133 ( a @xmath12 , low - luminosity and extremely distant fr - i radio source which is resolved by the evn into a core plus hot - spots , the latter being associated with the larger scale radio jet ) , vla j123642 + 621331 ( a dust enshrouded , optically faint
@xmath13 , @xmath14 starburst system waddington et al .
1999 ) and the faintest detection , vla j123646 + 621404 ( a face - on spiral galaxy at @xmath15 with a total evn flux density of @xmath16jy / beam ) .
the diversity of optical type is interesting but the real surprise is the detection of a radio - loud agn in the dust obscured , @xmath17 , starburst system .
this argues that at least some fraction of the optically faint radio source population harbour hidden agn ( this may be similar to the same obscured population detected by chandra ) .
these agn powered systems might be quite difficult to detect with scuba , if the dust temperatures are higher than that associated with pure star forming systems . in any case
, the detection of this system highlights the use of vlbi as a powerful diagnostic able to distinguish in principle ( via brightness temperature arguments ) between radio emission generated by nuclear starbursts and agn .
the evn observations of the hdf suggest that deep , wide - field vlbi studies are not only possible , but in principle they can deliver important astronomical results .
so far we have only scratched the surface . in a sense
, we are just beginning to appreciate the fact , that vlbi has reached a sensitivity level where we can expect to detect many discrete radio sources in a single field of view ( _ irrespective _ of where you point the telescopes ! ) .
this is quite a departure from the traditional vlbi `` postage stamp '' approach of observing specific , singular targets , often chosen from flux limited catalogues of the very brightest and most compact radio sources in the sky . in the short - term continuum vlbi sensitivity
is set to continue its steady improvement
three new telescopes are expected to come on - line ( including the 64-m sardinia radio telescope , see grueff these proceedings ) and various upgrades are already well advanced ( e.g. the replacement of the surface of the lovell 76-m ) .
recent evn test observations have demonstrated that the new 2-head , 512 mbit / sec data recording system can reach thermally limited r.m.s .
noise levels of @xmath18 microjy / beam with ( on - source ) integration times of only @xmath19 hours .
there is every reason to believe that much longer observations employing `` in - beam '' phase - calibration techniques ( rather than external phase - calibration ) can also attain thermally limited noise levels .
for example , currently a 24 hour observation of a carefully selected deep field should attain a r.m.s .
noise level of @xmath20 microjy / beam .
similar noise levels can be achieved with a global vlbi array . at these sort of flux density levels ,
the radio sky becomes so crowded that it makes sense to image out a large fraction of the primary beam of the largest individual antennas in the array .
the anticipated introduction of extremely fine spectral resolution and sub - second integration times at the evn data processor at jive , will make it possible to do just this .
dozens of faint sub - mjy and microjy radio sources will thus be simultaneously targeted and imaged with milliarcsecond resolution , full uv - coverage and microjy sensitivity .
the `` postage stamp '' era of vlbi will thus be consigned to the astronomical dust bin , and with it , the current limitations of vlbi to conduct unbiased , general sky surveys .
[ inbeam ] highlights the type of vlbi , deep field observing philosophy that might be employed .
the `` in - beam '' calibrator is chosen to be bright enough to provide adequate snr in the phase solutions but not so bright as to impose dynamic range limitations on the surrounding field ( see garrett 2000b for a more detailed explanation ) .
pc disk - based `` recording '' systems ( e.g. mkv , whitney 2001 and pc - evn , parsley 2001 ) are expected to replace the aging mkiv and vlba tape recorders over the next few years .
initially they provide the same level of economy , capacity and recording data rates as current tape based systems but with a much reduced level of maintenance and a much higher level of reliability .
in addition , the capacity of these disk - based systems is also expected to increase rapidly , as they take advantage of the continuing commercial development of pcs and their peripheral storage devices .
commercially operated optical fibre networks also appear capable of distributing vlbi data from telescopes to a central correlator in real - time .
tests are already being conducted , both within europe , the us and between the two continents ( see schilizzi these proceedings & schilizzi 2002 ) . in any case , the expectation is that vlbi will be able to utilise several ghz of bandwidth on the same time - scale as the evla and e - merlin begin operation i.e. before the end of the current decade .
employing several ghz of bandwidth has of course implications down the line : a more capable correlator will be required , perhaps not very different to the system now being designed for the evla and e - merlin .
vlbi data acquisition electronics will also need to be completely redesigned , and broad - band receivers must be introduced at telescopes across the network
. a new instrument will thus emerge ( currently dubbed @xmath2evn ) that will deliver substantial benefits : sub - microjy noise levels and almost full uv - coverage at @xmath21 cm ( see fig .
[ uvcov ] ) .
at these sub - microjy noise levels , it should be possible to begin to detect not only traditional vlbi targets such as low - luminosity agn , sn ( snr ) etc .
, but also cosmologically distant , nuclear starburst galaxies , such as those now observed ( but typically not well resolved ) by connected arrays in the hubble deep field . fig .
[ arp220 ] shows how the radio and sub - mm flux density of arp 220 changes with redshift . in nuclear starbursts ( such as arp 220 ) a large fraction of the radio emission
is contained within a region of intense star formation , usually less than a kpc across . at cosmological distances 1
kpc corresponds to an angular diameter of @xmath22 arcsecond on the sky , a scale size that is ideally suited to a combined @xmath2evn and e - merlin array . with a star formation rate ( sfr ) of @xmath23
m@xmath4/yr , fig .
[ arp220 ] shows that arp 220 could in principle be detected by the @xmath2evn out to cosmologically interesting redshifts . since many of the galaxies in the hdf have inferred sfr that are more than an order of magnitude larger than arp 220 , the radio structure of galaxies at even earlier epochs may be probed .
it is not clear what to expect but the view is expected to be dramatic , as massive star formation ( and its violent `` radio loud '' aftermath
sn , snr and grbs ) rampages unchecked through still - forming galaxies .
a detailed study of the radio structure of such distant systems will probably require the nanojy sensitivity of a high resolution , next generation radio instrument , such as the square km array ( ska ) .
deep radio continuum images can tell us how star formation is distributed through a given system , providing clues to the nature of the galaxy formation process itself and the importance of galaxy interactions and mergers , not to mention the inter - relationship between agn activity and star formation .
but in the meantime , the @xmath2evn with baselines on the scale of several thousand km , ghz of bandwidth , `` fantastic '' input and output data rates , and of course microjy sensitivity , can make a useful start in this area .
the @xmath2evn will also form a natural test - bed for the ska , encountering many of the same problems and initial limitations .
the next few years promise to be exciting and challenging , for radio astronomy in general but vlbi in particular .
several of the results shown here were made in collaboration with several colleagues .
i d particularly like to acknowledge contributions made by ger de bruyn , simon garrington and tom muxlow .
i d also like to thank the organisers of the meeting for their support , _ pazienza _ ( particularly with regard to the production of this paper ) and the memorable entertainment at the conference dinner !
9999 aussel , h. , cesarsky , c.j . ,
elbaz , d. , starck , j.l . , 1999 ,
a&a _ 342 _ , 313 barger , a.j . ,
cowie , l.l . ,
trentham , n. , et al . , 1998 ,
aj _ 117 _ , 102 condon , j.j .
, 1992 , ara&a , _ 30 _ , 575 garrett , m.a . ,
porcas , r.w .
, pedlar , a. et al . , 1999 ,
newar , _ 43 _ , 519 ( astro - ph/9906108 ) .
garrett , m.a . , de bruyn , a.g . ,
giroletti , m. , baan , w.a . ,
schilizzi , r.t .
, 2000a , a&a letters , _ 361 _ , l41 ( astro - ph/0008509 ) garrett , m.a .
, 2000b , _ procs .
of the vsop symposium
: h. hirabayashi , p.g .
edwards , and d.w .
murphy , published by the isas , pp 269 - 272 ( astro - ph/0003073 ) garrett , m.a . ,
muxlow , t.w.b . ,
garrington , s.t .
et al . , 2001 , a&a letters , _ 366 _ , l5 ( astro - ph/0008509 ) garrett , m.a . ,
2002 , a&a letters , 384 , l19 - 22 ( astro - ph/0202116 ) hughes , d.h . , serjeant , s. , dunlop , j. , et al .
, 1998 , nature , _ 394 _ , 241 muxlow , t.w.b .
, wilkinson , p.n . , richards , a.m.s . , et al . , 1999 , newar , _ 43 _ , 623 norris , r.p , hopkins , a. , sault , r.j . , et al .
, 2000 , _ procs . of the eso workshop deep fields _ ,
cristiani , renzini & williams , springer pp 135 - 138 parsley , s. , 2001 , `` pc - evn project plan '' , evn document 113 richards , e.a .
, kellermann , k.i . , fomalont , e.b . , et al . , 1998 , aj , _ 116 _ , 1039 richards , e.a . 2000 , apj , _ 533 _ , 611 schilizzi , r.t . , 2002 ,
`` networking in europe status and plans '' , _ proc .
e - vlbi workshop _ , web.haystack.mit.edu/e-vlbi/ waddington , i. , windhorst , r.a , cohen , et al .
, 1999 , apjl , _ 526 _ , l77 whitney , a.r . , 2001 ,
mark v memo series no .
1 williams , r.e . ,
blacker , b , dickinson , m. et al .
, 1996 , aj , _ 112 _ , 1335 | i review the current status of deep , wide - field vlbi continuum surveys .
i also discuss anticipated short and long - term improvements in sensitivity ( e.g. the eevn ) , and the science these developments will enable .
.5 cm |
let @xmath1 be a simple undirected graph and @xmath0 a subset of @xmath2 .
consider the following repetitive polling game . at round 0 the vertices of @xmath0
are colored white and the other vertices are colored black . at each round , each vertex @xmath3 is colored according to the following rule . if at round @xmath4 the vertex @xmath3 has more than half of its neighbors colored @xmath5 , then at round @xmath6 the vertex @xmath3 will be colored @xmath5 .
if at round @xmath4 the vertex @xmath3 has exactly half of its neighbors colored white and half of its neighbors colored black , then we say there is a tie . in this case
@xmath3 is colored at round @xmath6 by the same color it had at round @xmath4 .
( peleg considered other models for dealing with ties .
we will refer to these models in section [ remarks ] .
additional models and further study of this game may be found at @xcite , @xcite , @xcite , @xcite and @xcite . )
if there exists a finite @xmath4 so that at round @xmath4 all vertices in @xmath2 are white , then we say that @xmath0 is a _ dynamic monopoly _ , abbreviated _
dynamo_. in this paper we prove [ main ] for every natural number @xmath7 there exists a graph with more than @xmath7 vertices and with a dynamic monopoly of 18 vertices .
we shall use the following notation : if @xmath8 then @xmath9 denotes the set of neighbors of @xmath3 .
we call @xmath10 the _ degree _ of @xmath3 . for every @xmath11
we define @xmath12 as a function from @xmath2 to @xmath13 , so that @xmath14 if @xmath3 is white at round @xmath4 and @xmath15 if @xmath3 is black at this round .
we also define @xmath16 , @xmath17 , @xmath18 and @xmath19
( the small black circles are the vertices @xmath20 . ) ] [ jj ] let @xmath21 be the graph in figure 1 .
let @xmath22 and let @xmath23 and @xmath24 .
we construct a graph @xmath25 by duplicating @xmath7 times the vertices in @xmath26 .
that is , @xmath27 where @xmath28 \times d\ ] ] and @xmath29 \ } \nonumber \\ \cup \ { ( ( i , u),(i , v ) ) : ( u , v ) \in j , \ : u , v \in d , i \in [ n ] \ } \nonumber \end{aligned}\ ] ] ( here , as usual , @xmath30 $ ] denotes the set @xmath31 ) .
note that for reasons of symmetry , at a given round , all copies of a vertex in @xmath32 have the same color .
thus we may write `` @xmath33 is white at round 3 '' instead of `` @xmath34 is white at round 3 for every @xmath35 $ ] '' etc .
the following table describes the evolution of @xmath25 .
the symbol 1 stands for white and 0 stands for black .
note that the table does _ not _ depend on @xmath7 .
( this property is peculiar to the graph @xmath32 . in general graphs duplication of vertices may change the pattern of evolution of the graph ) .
@xmath36 the table shows that at round 20 the entire system is white and therefore @xmath0 is a dynamo .
the reader may go through the table by himself , but in order to facilitate the understanding of what happens in the table let us add some explanations as to the mechanism of `` conquest '' used in this graph .
we say that round @xmath37 _ dominates _ round @xmath38 if @xmath39 .
we shall make use of the following obvious fact : [ monotony ] if round @xmath37 dominates round @xmath38 ( @xmath40 ) then round @xmath41 dominates round @xmath42 . by applying this observation @xmath43 times
, we find that if round @xmath37 dominates round @xmath38 then round @xmath44 dominates round @xmath45 ( @xmath46 ) . by looking at the table
one can see that in the graph @xmath25 round 2 dominates round 0 and thus we have [ blinking ] round @xmath47 dominates round @xmath43 in @xmath25 for every @xmath48 we say that a vertex @xmath3 _ blinks _ at round @xmath4 if @xmath49 for every @xmath50 .
we say that a vertex @xmath3 is _ conquered _ at round @xmath4 if @xmath51 for every @xmath50 .
examining rounds @xmath52 to @xmath53 in the table and using corollary [ blinking ] one can see that @xmath54 and @xmath55 are conquered at round 0 , and in addition @xmath56 and @xmath33 are conquered at round 2 .
furthermore , every vertex in @xmath25 blinks either at round 1 or at round 2 .
finally , we have [ tieconquer ] if at round @xmath4 a vertex @xmath3 in @xmath25 has at least half of its neighbors conquered then @xmath3 is conquered at round @xmath57 . _
proof : _ every vertex in @xmath25 blinks either at round 1 or at round 2 , and hence @xmath3 is white either at round @xmath6 or at round @xmath57 . from this round on ,
at least half of the neighbors of @xmath3 are white , so @xmath3 will stay white .
@xmath58 now the vertices will be conquered in the following order : @xmath59 , @xmath60 , @xmath61 . eventually , the entire graph is colored white .
@xmath25 is a graph with @xmath62 vertices and @xmath0 is a dynamo of size 18 , proving theorem [ main ] .
the result of section [ proofmain ] gives rise to the following questions : does there exist an infinite graph with a finite dynamo ? the answer is _
no_. this follows from the following theorem : if @xmath0 is finite then @xmath63 is finite for all @xmath64 . moreover , every vertex in @xmath63 has a finite degree .
_ proof : _ the proof is by induction on @xmath4 . for @xmath65 the theorem is true because every vertex @xmath66 with an infinite degree becomes black at round 1 . for @xmath67 , if @xmath68 and @xmath3 has an infinite degree @xmath69 then by the induction hypotheses @xmath70 and @xmath71 .
hence @xmath72 and @xmath15 .
if @xmath73 has a finite degree then @xmath3 has a neighbor in @xmath74 . by the induction hypotheses
only finitely many vertices have such a neighbor , and thus @xmath63 is finite .
@xmath58 the next question deals with other models considered by peleg : [ 2 g ] do we still have a dynamo of size o(1 ) if we change the rules of dealing with ties ? ( e.g.
if a vertex becomes black whenever there is a tie . )
the answer here is _
yes_. if @xmath1 is a graph , introduce a new vertex @xmath75 for every @xmath8 and consider the graph @xmath76 where @xmath77 and @xmath78 if @xmath0 is a dynamo of @xmath79 according to the model in theorem [ main ] , then it is easy to prove that @xmath80 is a dynamo of @xmath81 .
but all vertices of @xmath81 have odd degrees , and thus ties are not possible and @xmath82 is a dynamo of @xmath81 according to _ any _ rule of dealing with ties .
therefore , for every @xmath83 the graph @xmath84 has a dynamo of size 36 .
let @xmath85 be a real number . consider the following model , which will henceforth be called _ the @xmath86-model_. at every round , for every vertex @xmath3 with @xmath87 neighbors colored black and @xmath88 neighbors colored white , if @xmath89 then @xmath3 is colored white at the next round , otherwise it is black . for the sake of simplicity
we will assume that @xmath86 is irrational and that there are no isolated vertices , so that @xmath90 is impossble .
the most interesting question regarding this model is whether there exist graphs with o(1 ) dynamo like in theorem [ main ] .
this question is as yet open .
we only have some partial results , which can be summarized as follows : 1 .
if @xmath86 is big enough then the size of a dynamo is @xmath91 .
2 . if @xmath86 is small enough then there exist graphs in which the size of a dynamo is @xmath92 .
if there exist graphs with o(1 ) dynamo then the number of rounds needed until the entire system becomes white is @xmath93 . for every @xmath96 ,
let @xmath97 be the set of edges with one vertex in @xmath98 and the other not in @xmath98 .
call @xmath99 .
note that @xmath100 is the set of vertices which are white at both round 0 and round 1 .
every @xmath101 is connected to at most @xmath102 vertices in @xmath103 and at most @xmath104 vertices outside of @xmath0 .
therefore we have @xmath105 let @xmath4 be fixed . by definition @xmath107 .
let @xmath108 , and let @xmath109 .
more than @xmath110 of the neighbors of @xmath3 are white at round @xmath4 and more than @xmath110 of the neighbors of @xmath3 are white at round @xmath111 . thus more than @xmath112 of the neighbors of @xmath3 belong to @xmath98 .
we therefore have @xmath113 which implies @xmath106 . by induction @xmath114 for all
if we begin with a dynamo then for some finite @xmath115 we have @xmath116 and @xmath117 @xmath58
let @xmath122 denote the sum of the degrees of the vertices in @xmath98 .
recall that every @xmath101 is white at both round 0 and round 1 , and thus @xmath123 and @xmath124 .
therefore , @xmath125 .
again , let @xmath4 be fixed , let @xmath126 be as in the proof of theorem [ rho3 ] and let @xmath109
. more than @xmath127 of the neighbors of @xmath3 are white at round @xmath4 and more than @xmath127 of the neighbors of @xmath3 are white at round @xmath111 . thus more than @xmath128
of the neighbors of @xmath3 belong to @xmath98 .
therefore , we have let @xmath136 be as defined in the answer to question [ 2 g ] .
construct @xmath137 by eliminating @xmath138 from @xmath136 and connecting @xmath139 to @xmath33 and @xmath140 ( but _ not _ to @xmath141 ) .
note that in @xmath137 the vertex @xmath141 is connected only to @xmath142 and to @xmath143 .
construct @xmath144 as in the construction of @xmath25 , where the duplicated vertices are all black vertices except for @xmath145 and @xmath146 .
( note that the graphs are constructed separately , namely , the sets of vertices of @xmath147 and @xmath148 are disjoint for @xmath149 . )
now connect the graphs in the following way .
first , eliminate the copies of @xmath150 from all graphs except for @xmath151 .
note that in @xmath147 there are @xmath152 copies of @xmath141 ( when @xmath153 ) . divide them into 32 disjoint sets @xmath154 , of size @xmath155 each .
now connect the vertices in @xmath156 to the copy of @xmath145 in @xmath157 , connect @xmath158 to the copy of @xmath146 , and connect each one of @xmath159 to a respective white vertex in @xmath157 ( see in figure 3 ) . .
the vertices under the numeral 1 are the 32 copies of @xmath141 in @xmath151 . under the numeral 2
are the 32 unduplicated vertices in @xmath160 ( @xmath145 , @xmath146 and the initiallly white vertices ) . under the numeral 3
are the 64 copies of @xmath141 in @xmath160 , under the numeral 4 are the 32 unduplicated vertices in @xmath161 , under the numeral 5 are the 128 copies of @xmath141 in @xmath161 , and so on . ] 1 .
all vertices of the obtained graph blink either at round 1 or at round 2 . 2 .
all vertices of @xmath162 are eventually conquered .
( the evolution of this conquest is similar to the one in theorem [ main ] . )
3 . if all copies of @xmath141 in @xmath147 are conquered at a certain round , then all vertices of @xmath157 are eventually conquered .
( again , the evolution is similar to the one in theorem [ main ] .
note that we need the bound @xmath163 in order to have @xmath145 and @xmath146 conquered . )
bermond and d. peleg , the power of small coalitions in graphs , _ proc .
2nd colloc .
on structural information and communication complexity _ , olympia , greece , june 1995 , carleton univ .
press , 173 - 184 . | the paper deals with a polling game on a graph . initially ,
each vertex is colored white or black . at each round , each vertex is colored by the color shared by the majority of vertices in its neighborhood , at the previous round .
( all recolorings are done simultaneously ) .
we say that a set @xmath0 of vertices is a _ dynamic monopoly _ or _ dynamo _ if starting the game with the vertices of @xmath0 colored white , the entire system is white after a finite number of rounds .
peleg @xcite asked how small a dynamic monopoly may be as a function of the number of vertices .
we show that the answer is o(1 ) . |
* part i * : we first show that eq .
( 1 ) is a legitimate transformation describing the cv quantum teleportation that leads to exactly the same mapping in the schr " odinger picture [ 22 ] . in ref .
[ 21 ] , eq .
( 1 ) was used for the study of teleporting temporal correlation functions by only noting that the operator @xmath136 acts like a classical number due to @xmath137=0 $ ] .
in contrast , we rigorously justify eq .
( 1 ) as follows . note that @xmath138 can be understood as superposing two modes @xmath13 and @xmath15 at a 50:50 beam splitter followed by the nondegenerate parametric amplification ( gain @xmath139 ) with the idler mode @xmath14 .
this suggests the conjugate evolutions for @xmath14 and @xmath140 , @xmath141 with the heisenberg transformations in eqs .
( 1 ) and ( 8) taken as @xmath142 ( @xmath17 , where @xmath31 is a global unitary operator corresponding to cv teleportation , one can obtain the characteristic function of the output , @xmath143 , where @xmath144 , as @xmath145 which is in agreement with [ 22 ] . *
part ii * : we next prove that there exist quantum channels that do not have the usual epr correlation , i.e. @xmath45 , but can teleport any input state with fidelity arbitrarily close to unity . for simplicity ,
let us consider a class of entangled channels produced by injecting identical fields ( with only a phase shift to the second ) of modes @xmath146 and @xmath147 to a 50:50 beam splitter .
the output modes @xmath148 and @xmath149 give the fidelity in eq .
( 2 ) as @xmath150 . for high fidelity , we only need to maximize @xmath151 for a single mode ; @xmath152 can be similarly maximized by applying @xmath153 phase - shift ( @xmath154 ) .
to ensure the no - epr - correlation , @xmath45 , for this channel , we impose the condition @xmath155 ( no squeezing ) .
our task is now to make @xmath156 close to unity with @xmath155 .
obviously , the quadrature distribution @xmath157 must be sharply centered around @xmath158 to obtain a large @xmath159 , but presumably with a rather long tail to simultaneously ensure a finite variance , @xmath160 .
this leads us to envision a probability distribution that decreases with @xmath161 more rapidly than the gaussian function .
we can construct a number of such distributions analytically , e.g. @xmath162 ( @xmath163 ) .
after normalization , we obtain a class of probability distributions @xmath164 and find that @xmath165 approaches unity as @xmath166 increases .
one can construct a corresponding non - gaussian entangled state , @xmath167 ( @xmath168 : beam - splitter ) with two single - mode states @xmath169 and @xmath170 , where @xmath171 and @xmath172 are @xmath0- and @xmath1-quadrature eigenstates with the probability distribution @xmath173 in eq .
the uncertainty principle yields @xmath174 for the state @xmath175 .
no squeezing is then assured for arbitrary quadratures @xmath176 with a little algebra .
furthermore , it can be shown that the state @xmath177 asymptotically gives the unit fidelity for all input states ; its characteristic function @xmath178 , where @xmath179 , becomes very flat over the entire range of @xmath180 with @xmath166 increased , which ensures a high - fidelity via eq . ( 9 ) . as an example , in fig . 3
, we plot the wigner function of the teleported state for the input fock state @xmath181 .
the overall features are successfully teleported except for a relatively long tail in the output , whose effect on fidelity is however negligible . *
part iii * : finally , we show how the separability condition in ( 3 ) can be derived employing an arbitrary function @xmath49 . it is well known that for a separable state , the density operator under partial transposition ( pt ) is still positive - semidefinite [ 7 ] , thus satisfying all requirements imposed on a legitimate quantum state . as demonstrated below , this leads to an upper / lower bound of the value @xmath51 that a separable state can take , where @xmath182 . for each @xmath49
, one can find out such bounds as prescribed below , the violation of which is a sufficient condition for entanglement . first , note that when @xmath14 and @xmath15 are input modes to a 50:50 beam splitter [ fig . 1 ( a ) ] , the operators @xmath183 and @xmath184 correspond to two commuting observables @xmath185 and @xmath186 , respectively , at the output modes . from the viewpoint of the output state , it is thus obvious that the hermitian operator @xmath50 , for any function @xmath49 , can be regarded as a proper symmetric - ordered operator .
therefore , one can generally express the quantum average @xmath51 by means of two - mode wigner distribution @xmath187 as @xmath188,\nonumber\\ \label{eqn : exp}\end{aligned}\ ] ] where the epr operator @xmath189 is replaced by the corresponding classical amplitudes ( @xmath190/@xmath191 : real / imaginary parts ) . under pt ,
the wigner function changes as @xmath192 [ 8 ] and eq .
( [ eqn : exp ] ) becomes @xmath193 .
\nonumber\\ \label{eqn : bound1}\end{aligned}\ ] ] with the change of integration variables ( redefining new modes ) as @xmath194 and @xmath195 , the above expression can be recast to @xmath196 where @xmath197 represents a certain two - mode wigner function .
we now see that the extremal values of @xmath198 for _ all _ quantum states are the same as those of a single - mode @xmath199 as @xmath200 now involves only one mode @xmath201 in eq .
( [ eqn : bound11 ] ) . from now on ,
we thus work with only a single - mode state @xmath202 having a wigner function @xmath203 . writing @xmath203 as the fourier transform of the characteristic function @xmath204 $ ] , i.e. , @xmath205 , we obtain @xmath206 , \label{eqn : bound2}\end{aligned}\ ] ] where @xmath207 now the problem is reduced to that of obtaining extremal expectation values of the operator @xmath208 in eq .
( [ eqn : bound2 ] ) . obviously , the maximum ( minimum ) value of @xmath209 $ ] is the maximum ( minimum ) eigenvalue of @xmath208 .
one can show by a direct calculation that the operator @xmath208 does not have off - diagonal elements in the number - state basis , i.e. @xmath210 for @xmath211 .
thus , the eigenvalues of @xmath208 are the diagonal elements @xmath212 given by @xmath213 using @xmath214 .
( @xmath56 : laguerre polynomial , @xmath57 : bessel function ) the ensemble average @xmath215 in eq .
( [ eqn : bound2 ] ) can thus take a value only in the range @xmath216 $ ] where @xmath217 over @xmath218 .
therefore , for any function @xmath49 , a separable state must satisfy @xmath52 as an example , our method , applied to the lowest - order epr correlation @xmath219 , recovers duan _ et al .
_ s criterion . for this case ,
the argument of @xmath49 in eq .
( [ eqn : bound1 ] ) becomes @xmath220 @xmath221 .
then , by redefining a single - mode @xmath222 , we similarly obtain the bounds , eq .
( [ eqn : bound33 ] ) , with only the replacement @xmath223 $ ] .
identifying @xmath224 , one obtains the minimum @xmath225 with @xmath90 .
therefore , we obtain the duan _
s criterion as @xmath226 . | we present a formalism to derive entanglement criteria beyond the gaussian regime that can be readily tested by only homodyne detection .
the measured observable is the einstein - podolsky - rosen ( epr ) correlation .
its arbitrary functional form enables us to detect non - gaussian entanglement even when an entanglement test based on second - order moments fails .
we illustrate the power of our experimentally friendly criteria for a broad class of non - gaussian states under realistic conditions .
we also show rigorously that quantum teleportation for continuous variables employs a specific functional form of epr correlation .
quantum entanglement plays a key role in making quantum - mechanical predictions distinct from their classical counterparts .
its characterization and detection have been a topic of crucial importance that has attracted a great deal of effort , both theoretically and experimentally . despite remarkable progress
, there still exists a pressing demand for further developments particularly for continuous variables ( cvs ) . in the cv regime ,
gaussian entangled states have been the subject of study , but now considerable attention has been directed to non - gaussian states .
this may be attributed to the fact that non - gaussian entangled states not only provide practical advantages over their gaussian counterparts , e.g. in quantum teleportation @xcite and dense coding @xcite , but also turn out to be essential , e.g. for universal quantum computation @xcite and nonlocality test @xcite .
the quantum state of a cv system is fully described in phase space by representing two quadrature variables @xmath0 ( position ) and @xmath1 ( momentum ) . to address the correlation of two cv systems ,
the epr operators @xmath2 and @xmath3 can be defined , as einstein , podolsky , and rosen ( epr ) originally envisioned @xcite . here
@xmath4 is an arbitrary real number and @xmath5 and @xmath6 label each subsystem . when the sum of the variances , @xmath7 , is smaller than @xmath8 , the state of the total system is inseparable @xcite and such states are said to be epr - correlated .
in fact , it has been shown that the epr correlation , after local unitary operations , is a necessary and sufficient condition for two - mode gaussian entanglement @xcite ( in this letter , we study a simple case , where we use @xmath9 instead of @xmath10 assuming @xmath11 ) . using the uncertainty principle as a requirement of a physical system in conjunction with the partial transposition @xcite , simon found a necessary and sufficient condition for gaussian entangled states ( simon criterion ) @xcite ; this criterion is also concerned with the second moments of the quadrature variables @xcite .
these criteria are theoretically easy to calculate and readily tested experimentally since the quadrature variables can be measured using highly - efficient homodyne detectors . beyond the gaussian regime ,
some entanglement criteria have also been proposed @xcite .
shchukin and vogel particularly derived a hierarchy of entanglement conditions using higher - order moments @xcite . for finite - dimensional systems , a remarkable criterion based on covariance matrices that unifies other criteria as its corollaries has been developed @xcite .
however , these tools are not necessarily practical for infinite - dimensional cv states because they may be nontrivial to assess theoretically and implement experimentally @xcite .
some criteria are at least theoretically easier to assess but they do not seem useful for a large class of non - gaussian states . therefore , new entanglement criteria beyond the gaussian regime must be developed , desirably in an experimentally friendly form , which is the objective of the current work .
here we present a formalism employing the usual epr operator @xmath12 in an arbitrary functional form , which is readily testable with highly efficient homodyners and is theoretically convenient to calculate .
these functional epr criteria turn out to detect cv entanglement well beyond the simon criterion , thereby proving genuine non - gaussian entanglement .
our criterion can be efficiently tested under realistic conditions including environmental interactions and detector inefficiency .
we illustrate the power of our formalism by considering a broad class of non - gaussian states and show that it can be a crucial tool in addressing cv problems of fundamental importance , e.g. the robustness of gaussian versus non - gaussian entangled states against decoherence . in order to present the new entanglement criterion
, we briefly revisit the so - called braunstein - kimble ( bk ) protocol which is the standard for cv teleportation @xcite . unlike the original description , we study the protocol in the heisenberg picture in order to gain a crucial insight .
let @xmath13 be an unknown input which is to be teleported , and @xmath14 and @xmath15 the entangled quantum channel sent to alice and bob , respectively .
each field can be decomposed into real and imaginary parts ( two quadrature amplitudes ) , @xmath16 ( @xmath17 . in the bk protocol , alice first superposes her mode with the input at a 50:50 beam splitter to obtain two output modes , @xmath18 , and measures the @xmath0 ( @xmath1 ) quadrature of mode @xmath19 ( @xmath20 ) .
the measurement outcomes @xmath21 are transmitted to bob who performs the displacement @xmath22 with @xmath23 @xcite .
although @xmath24 and @xmath25 are classical numbers , if formally represented by quantum operators @xcite , bob s output becomes @xmath26 we rigorously justify eq .
( [ eqn : ht1 ] ) by showing that it leads to exactly the same mapping in the schr " odinger picture @xcite ( supplemental material ) . _
teleportation fidelity_as a figure of merit to assess the performance of the bk protocol , the output fidelity @xmath27 averaged over all input coherent states @xmath28 ( @xmath29 : amplitude ) can be used . noting that the mean amplitudes of input and output fields always agree in the bk protocol @xcite
, we realize that it suffices to look into the fidelity for the vacuum state input @xmath30 .
let @xmath31 be the global unitary operator that gives the output state @xmath32 after teleportation via a quantum channel @xmath33 ( supplemental material ) .
the fidelity is given by @xmath34 with the normal - ordered expansion of the vacuum @xmath35 @xcite .
using eq .
( [ eqn : ht1 ] ) and the vanishing contribution of the input vacuum state under normal - ordering , the output fidelity is reduced to an average over the quantum channel @xmath33 , @xmath36 using the identity @xmath37 @xcite .
we see now that all orders of epr correlations @xmath38 contribute to the fidelity nontrivially , and this can be used to address various issues on quantum teleportation using non - gaussian channels . as an example , due to an inequality @xmath39
, we deduce @xmath40 .
therefore , the condition @xmath41 alone , using gaussian or non - gaussian channels , guarantees the fidelity over the classical limit @xmath42 @xcite . for a gaussian channel ,
higher - order correlations are constrained by the lowest one @xmath43 , and @xmath44 becomes a necessary condition for faithful teleportation .
surprisingly , using the formulation above , we find that for a non - gaussian channel , it is possible to have near - unit fidelity even without the usual epr correlation , @xmath45 .
therefore , the epr correlation is not a necessary condition for a successful teleportation in the non - gaussian regime .
remarkably , beyond the teleportation context , this suggests _ a possibility of detecting genuine non - gaussian entanglement _ , for which the simon criterion fails .
note that ( i ) the fidelity above 1/2 proves quantum entanglement @xcite and ( ii ) the fidelity can be predicted also by directly measuring @xmath46 of the quantum channel over @xmath47 runs , as @xmath48 .
( see also _ implementation _ below and fig . 1 ( a ) .
) _ entanglement criteria_we present a formalism to construct an arbitrary functional form of epr correlation as an entanglement criterion .
given a general function @xmath49 , we consider an ensemble average of the hermitian operator @xmath50 , with @xmath12 .
as detailed in supplemental material , for separable states , the value of @xmath51 is bounded , @xmath52 here the upper ( lower ) bound for separability is given by @xmath53 over @xmath54 , where @xmath55 ( @xmath56 : laguerre polynomial , @xmath57 : bessel function ) .
eq .
( [ eqn : bound3 ] ) is readily integrated for a regular function @xmath49 . if the inequality ( [ eqn : criterion ] ) is violated for any function @xmath49 , the state is entangled .
our method , when applied to the lowest - order epr correlation @xmath58 , recovers duan _ et al .
_ s criterion . for higher - order criteria , using @xmath59
, we obtain separable conditions @xmath60 . below
we adopt a general @xmath49 in the taylor - series form and demonstrate its usefulness in detecting cv entangled states broadly . _
implementation_separable states must satisfy the inequality ( [ eqn : criterion ] ) also under any local unitary transformations .
this reasoning leads to a comprehensive set of entanglement criteria based on the epr operator using generalized orthogonal quadratures .
that is , the quadrature amplitudes under rotation , @xmath61 and @xmath62 @xmath63 can be employed and the violation of ( [ eqn : criterion ] ) for any single pair @xmath64 is sufficient to verify entanglement . given an entangled state , the violation usually occurs over a range of values @xmath65 and @xmath66 for a fixed @xmath49 , making an experimental test robust . in experiments ,
the generalized @xmath67 can be measured by first phase - shifting each mode with @xmath68 @xmath63 and injecting them to a 50:50 beam - splitter ( fig .
1 ( a ) ) .
one then measures the fixed quadrature @xmath69 and @xmath70 at each output , respectively , using homodyne detectors .
( @xmath71 and @xmath72 ) .
we emphasize that neither a full state tomography nor the reconstruction of a probability distribution is needed for this test of entanglement : the epr value @xmath73 obtained at each run ( @xmath74 ) is plugged into a test function @xmath49 to construct the average over @xmath47 runs as @xmath75 .
it is then compared with the bounds in eq .
( [ eqn : bound3 ] ) to verify entanglement .
as one obtains a finite number @xmath47 of data , it is of practical importance to see whether such a finite number of data suffices to beat the separability bounds properly .
this issue can be addressed by looking into the statistical error @xmath76 of the mean , @xmath77 , where @xmath78 we include this analysis below to demonstrate the power of our criteria under realistic conditions .
_ trial functions_for an entanglement test , we employ a function @xmath79 , an exponentially decreasing function multiplied by a finite polynomial .
this form can be regarded as the most general in view of the taylor - series expansion of a regular function ( @xmath80 ) .
our criterion becomes more powerful by adopting higher - order polynomials ; given an entangled state , our task is to find the set of parameters @xmath81 for which the inequality ( [ eqn : criterion ] ) is violated .
if it occurs at a level @xmath82 , so does it at higher levels @xmath83 .
the same logic also applies to the degree of violation measured by , e.g. , @xmath84 optimized over the parameters @xmath81 .
that is , the larger the @xmath82 value , the higher the confidence level of the test for a given number @xmath47 of homodyne measurements [ fig . 1 ( d ) ] .
in this letter we illustrate our method by confining the polynomial to the first order , @xmath85 .
remarkably , a broad class of non - gaussian entangled states of current theoretical and experimental relevance can be detected with it . using @xmath85 , eq .
( [ eqn : bound3 ] ) gives @xmath86 , \label{eqn : bound30}\end{aligned}\ ] ] whose maximum is relevant for entanglement detection .
* case ( i ) * @xmath87 : max @xmath88 is @xmath89 when @xmath90 .
therefore , if there exists any @xmath91 such that @xmath92 , the state is entangled .
for example , with @xmath93 , the separability bound becomes 1/2 , which is known as the classical limit for teleportation @xcite .
* case ( ii ) * @xmath94 : in this case , the number @xmath95 yielding maximum @xmath96 in eq . ( [ eqn : bound30 ] ) depends on @xmath97 and @xmath98 .
it should thus be obtained case by case , which is still straightforward to calculate . for a given state ,
the violation of ( [ eqn : criterion ] ) usually occurs in a broad range of @xmath97 and @xmath98 , which makes our scheme robust against noise .
below we consider possible experimental imperfections like channel loss or thermal noise during signal transmission and detector inefficiency or electronic noise in homodyne detection .
these errors can be modeled by @xmath99 where @xmath100 is the mode actually detected , @xmath101 the original mode , and @xmath102 the vacuum or thermal mode characterizing noise .
the results below are all obtained analytically .
our method is also compared with other cv criteria directly testable by homodyne detection such as the simon criterion @xcite and the entropic criterion @xcite .
_ examples_let us first consider a class of dephased cat - states @xcite , @xmath103 , where @xmath104 is a coherent state of amplitude @xmath105 and @xmath106 represents the degree of dephasing .
they are entangled for all @xmath107 and @xmath108 , which is not detected by the simon criterion .
the entropic criterion detects entanglement only for a large @xmath105 and @xmath109 , despite the improvement with the generalized renyi entropy @xcite .
in contrast , our criterion detects entanglement for any @xmath107 and @xmath108 , with @xmath110 in fig . 1 ( a ) . considering further the overall efficiency @xmath111 characterizing channel loss and detector efficiency , we plot the value of @xmath112 in fig . 1 ( b ) , where each point is optimized over the parameters @xmath113 $ ] and @xmath114 $ ] for a highly dephased state ( @xmath115 ) .
the positive values in _ all _ ranges imply the detection of non - gaussian entanglement with the degree of violation increased with @xmath105 and @xmath111 .
moreover , let us address a realistic situation where a total number @xmath47 of data is obtained from homodyne detection with detector efficiency @xmath111 and electronic noise , which can be described by mixing signal with a thermal state ( @xmath116 : thermal excitation number ) @xcite . in fig .
1 ( c ) , the case of dephased cat ( @xmath117 ) is plotted as a function of @xmath105 for @xmath118 and @xmath119 ( electronic noise ) , with the error bar .
the positive values above zero clearly verify entanglement .
secondly , let us consider a state @xmath120 .
we have checked that the simon criterion and our method detect entanglement for any @xmath121 and @xmath122 values , whereas the entropic method @xcite does so for only @xmath123 . in fig . 1 ( d ) , we plot @xmath84 for the state @xmath124 , which is improved by changing the test function from @xmath125 to @xmath126 .
furthermore , suppose that the state undergoes decoherence through a thermal reservoir .
the reservoir interaction is described by eq .
( [ eqn : emodel ] ) .
as @xmath127 gets larger the system becomes more decohered .
as shown in fig . 2 ( a ) , entanglement detection is significantly improved over the simon method ( square ) , particularly by modifying the test function from @xmath128 ( circle ) to @xmath85 ( triangle ) .
similar results are also obtained for higher - dimensional entangled states . finally , the classes of entangled states generated by practically available non - gaussian operations , i.e. , photon - subtraction @xcite and addition @xcite are of current interest @xcite . when these operations are applied to a two - mode squeezed state @xmath129 , the simon criterion fails to verify the entanglement of the photon - added state @xmath130 for @xmath131 , whereas our method detects entanglement for the whole range of @xmath132 .
furthermore , we address a question of fundamental importance raised recently by allegra _
et al . _ ,
i.e. , whether gaussian entanglement is more robust than non - gaussian entanglement against decoherence due to the thermal reservoir @xcite .
we show that non - gaussian entanglement can be more robust than gaussian entanglement in the parameter regime they studied @xcite .
the conclusion of @xcite was drawn by using several previously known criteria for verifying non - gaussian entanglement , which however did not outperform the simon criterion .
in contrast , our method analytically shows that the photon - subtracted states ( triangle : our criterion , square : simon criterion ) survive longer under thermal noise than their gaussian counterparts of the same initial energy ( circle ) [ fig . 2 ( b ) ] .
similar results are also obtained for the photon added states . _
remarks_we have developed a method to derive entanglement criteria beyond the gaussian limit that can be efficiently tested by homodyne detection .
we illustrated our method by adopting two trial functions which broadly detect non - gaussian entanglements of fundamental and practical interests .
we also showed that our method is robust against experimental imperfections .
the trial functions can be made more powerful by incorporating higher - order polynomials , which do not however incur any extra experimental efforts .
on another side , the question on how much useful ( robust ) non - gaussian entangled states are under practical conditions is of importance for quantum information processing .
our study for quantum teleportation , clarified here as employing the epr correlation in a specific form , shows that non - gaussian channels can be more robust against decoherence than gaussian ones .
h. n. acknowledges the support by the nprp 4 - 520 - 1 - 083 from qatar national research fund and m. s. k. the financial support by the uk epsrc .
t. opatrny , g. kurizki , and d .- g .
welsch , , 032302 ( 2000 ) ; p. t. cochrane , t. c. ralph , and g. j. milburn , _ ibid .
_ * 65 * , 062306 ( 2002 ) ; s. olivares , m. g. a. paris , and r. bonifacio , _ ibid . _ * 67 * , 032314 ( 2003 ) ; a. kitagawa _
et al .
_ , _ ibid .
_ * 73 * , 042310 ( 2006 ) ; y. yang and f. l. li , _ ibid .
_ * 80 * , 022315 ( 2009 ) ; f. dellanno _ et al . _ , _ ibid . _ * 76 * , 022301 ( 2007 ) ; f. dellanno , s. de siena , and f. illuminati , _ ibid . _ * 81 * , 012333 ( 2010 ) ; s .- y .
lee _
et al . _ ,
phys .
rev . a 84 , 012302 ( 2011 ) .
a. kitagawa _
et al . _ , , 022334 ( 2005 ) .
s. lloyd and s. l. braunstein , , 1784 ( 1999 ) ; s. d. bartlett and b. sanders , _ ibid .
_ , * 89 * , 207903 ( 2002 ) . h. nha and h. j. carmichael , , 020401 ( 2004 ) ; r. garcia - patron _ et al .
_ , _ ibid .
_ * 93 * , 130409 ( 2004 ) .
a. einstein , b. podolsky , and n. rosen , phys . rev . * 47 * , 777 ( 1935 ) .
l .-
m .
duan _
et al . _ , ,
2722 ( 2000 ) .
a. peres , , 1413 ( 1996 ) .
r. simon , , 2726 ( 2000 ) .
g. giedke _
et al . _ , , 167904 ( 2001 ) .
m. hillery and m. s. zubairy , , 050503 ( 2006 ) .
g. s. agarwal and a. biswas , new j. phys .
* 7 * , 211 ( 2005 ) .
h. nha and j. kim , , 012317 ( 2006 ) : h. nha , , 014305 ( 2007 ) .
e. shchukin and w. vogel , , 230502 ( 2005 ) .
see also an uncertainty - relation - based approach .
h. nha and m. s. zubairy , , 130402 ( 2008 ) ; q. sun , h. nha , and m. s. zubairy , , 020101 ( 2009 ) .
o. ghne _
et al . _ , , 130504 ( 2007 ) ; o. gittsovich _
et al .
_ , , 052319 ( 2008 ) . s. p. walborn _
et al .
_ , , 160505 ( 2009 ) .
a. saboia , f. toscano , and s. p. walborn , , 032307 ( 2011 ) .
e. shchukin and w. vogel , , 200403 ( 2006 ) .
this proposed a homodyne correlation measurement that can be potentially useful for higher - order moments , for which the device becomes increasingly large .
s. l. braunstein and h. j. kimble , , 869 ( 1998 ) .
we here assume that both alice s and bob s modes have zero means , @xmath133 , e.g. the case of a usual two - mode squeezed state serving as a quantum channel . if not , bob can adjust the required displacement to @xmath134 to satisfy the condition @xmath135 . c. noh _ et al .
_ , , 230501 ( 2009 ) .
p. marian and t. a. marian , , 042306 ( 2006 ) .
s. m. barnett and p. m. radmore , _ methods in theoretical quantum optics _ , oxford university press ( 2003 ) .
s. l. braunstein , c. a. fuchs , and h. j. kimble , j. mod . opt . * 47 * , 267 ( 2000 ) ; k. hammerer _ et al .
_ , , 150503 ( 2005 ) .
a. ourjoumtsev _
et al . _ ,
nature physics * 5 * , 189 ( 2009 ) .
j. lodewyck _
et al . _ , , 042305 ( 2007 ) .
a. zavatta , s. viciani , and m. bellini , science * 306 * , 660 ( 2004 ) ; g. s. agarwal and k. tara , , 492 ( 1991 ) .
m. s. kim , j. phys .
b * 41 * , 133001 ( 2008 ) .
m. allegra , p. giorda , and m. g. a. paris , , 100503 ( 2010 ) .
see also k. k. sabapathy , j. s. ivan , and r. simon , , 130501 ( 2011 ) ; j. lee , m. s. kim , and h. nha , , 238901 ( 2011 ) . |
for biological reasons our perception of the universe is based on the observation of photons , most trivially by staring at the night - sky with our bare eyes .
conventional astronomy covers many orders of magnitude in photon wavelengths , from @xmath1 cm radio - waves to @xmath2 cm gamma rays of gev energy .
this 60 octave span in photon frequency allows for a dramatic expansion of our observational capacity beyond the approximately one octave perceivable by the human eye .
photons are not , however , the only messengers of astrophysical processes ; we can also observe cosmic rays and neutrinos ( and maybe also gravitons in the not - so - distant future ) .
particle astronomy may be feasible for neutral particles or possibly charged particles with energies high enough to render their trajectories magnetically rigid .
this new astronomy can probe the extreme high energy behavior of distant sources and perhaps provide a window into optically opaque regions of the universe . in these lectures
we will focus attention on the highest energy particles ever observed .
these ultrahigh energy cosmic rays ( uhecrs ) carry about seven orders of magnitude more energy than the lhc beam .
we will first discuss the requirements for cosmic ray acceleration and propagation in the intergalactic space .
after that we will discuss the properties of the particle cascades initiated when uhecrs interact in the atmosphere .
finally we outline strategies to search for physics beyond the highly successful but conceptually incomplete standard model ( sm ) of weak , electromagnetic , and strong interactions . before proceeding ,
we pause to present our notation . unless otherwise stated ,
we work with natural ( particle physicist s ) heaviside - lorentz ( hl ) units with @xmath3 the fine structure constant is @xmath4 .
all si units can then be expressed in electron volt ( ev ) , namely @xmath5 electromagnetic processes in astrophysical environments are often described in terms of gauss ( g ) units . for a comparison of formulas used in the literature we note the conversion , @xmath6 , @xmath7 and @xmath8 . to avoid confusion we will present most of the formulas in terms of ` invariant ' quantities , _
@xmath9 , @xmath10 and the fine - structure constant @xmath11 .
distances are generally measured in mpc @xmath12 .
extreme energies are sometimes expressed in eev @xmath13 ev .
the symbols and units of most common quantities are summarized in table [ tienda_de_cafe ] .
.symbols and units of most common quantities .
[ tienda_de_cafe ] [ cols="^,^,^",options="header " , ] consider first a flux of earth - skimming @xmath14 neutrinos with energy in the range @xmath15 .
the neutrinos can convert to @xmath14 leptons in the earth via the charged current interaction @xmath16 . in the ( perturbative ) sm
, the interaction path length for the neutrino is @xmath17 ^{-1 } \ , \ ] ] where @xmath18 is the charged current cross section for a neutrino energy @xmath19 .
the density of the material through which the neutrinos pass , @xmath20 , is about @xmath21 for the earth s crust . here
we have neglected neutral current interactions , which at these energies only reduce the neutrino energy by approximately 20% , which is within the systematic uncertainty . for @xmath22 , @xmath23 km .
let us assume some hypothetical non - perturbative physics process enhances the @xmath24 cross section .
then the interaction path length becomes @xmath25 ^{-1 } \ , \ ] ] where @xmath26 is the non - perturbative contribution to the cross section for @xmath19 .
once a @xmath14 is produced by a cc interaction , it can be absorbed in the earth or escape and possibly decay , generating a detectable air shower . at these high energies ,
the @xmath14 propagation length in the earth is dominated by energy loss rather than the finite @xmath14 lifetime .
the energy loss can be expressed as @xmath27 where @xmath28 characterizes energy loss due to ionization and @xmath29 characterizes losses through bremsstrahlung , pair production and hadronic interactions . at these energies , energy losses due to ionization turn out to be negligible , while @xmath30 @xmath31 @xcite . from eq . , we observe that the maximum path length for a detectable @xmath14 can be written @xmath32 where @xmath33 is the energy at which the @xmath14 is created , and @xmath34 is the minimal energy at which a @xmath14 can produce a shower big enough to be detected . for @xmath35 , @xmath36 km . the probability for a neutrino with incident nadir angle @xmath37 to emerge as a detectable @xmath14 is @xmath38 , \label{p}\ ] ] where @xmath39 is the chord length of the intersection of the neutrino s trajectory with the earth , with @xmath40 km the earth s radius .
note that we have neglected the possibility that non - perturbative processes could lead to a detectable signal , since the hadrons which dominate the final state will be absorbed in the earth .
the step function in eq .
reflects the fact that a @xmath14 will only escape the earth if @xmath41 , as illustrated in fig [ f : event ] . .
in the figure , the neutrino produces a lepton @xmath42 after traveling some distance @xmath43 inside the earth s crust .
if @xmath41 , the lepton will escape the earth and can generate an air shower . from ref .
@xcite.,scaledwidth=90.0% ] assuming an isotropic tau neutrino flux , the number of taus that emerge from the earth with sufficient energy to be detected is proportional to an `` effective solid angle '' @xmath44 evaluation of the integrals @xcite yields the unfortunate expression @xmath45
\left [ \left ( \frac{l_{\rm tot}^{\nu}}{2 r_{\oplus } } \right)^2 - \left ( \frac{l_{\rm tot}^{\nu}}{2 r_{\oplus } } + \left ( \frac{l_{\rm tot}^{\nu}}{2 r_{\oplus } } \right)^2 \right ) e^{-2r_{\oplus } / l_{\rm tot}^{\nu } } \right ] \ . \label{omegaeff}\end{aligned}\ ] ] at the relevant energies , however , the neutrino interaction length satisfies @xmath46 .
in addition , if the hypothesized non - perturbative cross section enhancement is less than typical hadronic cross sections , we have @xmath47 . with these approximations ,
. simplifies to @xcite @xmath48 equation describes the functional dependence of the earth - skimming event rate on the non - perturbative cross section .
this rate is , of course , also proportional to the neutrino flux @xmath49 at @xmath50 .
thus , the number of earth - skimming neutrinos is given by @xmath51 where @xmath52 is the number of earth - skimming events expected for some benchmark flux @xmath53 in the absence of new physics .
in contrast to eq . , the rate for quasi - horizontal showers has the form @xmath54 where @xmath55 is the number of quasi - horizontal events expected for flux @xmath53 .
now , without loss of generality , we normalize the neutrino flux to the waxmann - bahcall bound , _
i.e. _ @xmath56 .
we use the expected rates for the the benchmark flux to determine the values of @xmath52 and @xmath55 in eqs . and , ( @xmath57 and @xmath58 , as shown in table [ t : rates ] ) .
given a flux @xmath49 and new non - perturbative physics cross section @xmath26 , both @xmath59 and @xmath60 are determined . on the other hand ,
given just a quasi - horizontal event rate @xmath60 , it is impossible to differentiate between an enhancement of the cross section due to non - perturbative physics and an increase of the flux .
however , in the region where significant event rates are expected , the contours of @xmath60 and @xmath59 , given by eqs . and , are more or less orthogonal and provide complementary information .
this is illustrated in fig .
[ f : esqh ]
. plane .
note that the contours are roughly orthogonal , and so the two types of event provide complementary information about flux and cross section.,scaledwidth=99.0% ] with measurements of @xmath61 and @xmath62 , both @xmath63 and @xmath49 may be determined independently , and neutrino interactions beyond the ( perturbative ) sm may be unambiguously identified .
we now turn to determining the projected sensitivity of auger to neutrino fluxes and cross sections .
the quantities @xmath59 and @xmath60 as defined in eqs .
( [ es ] ) and ( [ qh ] ) can be regarded as the theoretical values of these events , corresponding to different points in the @xmath64 parameter space . for a given set of observed rates @xmath65 and @xmath66 ,
two curves are obtained in the two - dimensional parameter space by setting @xmath67 and @xmath68 .
these curves intersect at a point , yielding the most probable values of flux and cross section for the given observations .
fluctuations about this point define contours of constant @xmath69 in an approximation to a multi - poisson likelihood analysis @xcite .
the contours are defined by @xmath70 + 2\ , n_i^{\rm obs}\ , \ln \left [ n_i^{\rm obs}/ n_i \right ] \,\ , , \label{baker}\ ] ]
where @xmath71 es , qh @xcite . in fig .
[ statistics ] , we show results for three representative cases . assuming ( @xmath72 , @xmath73 ) , ( @xmath74 , @xmath75 ) , and ( @xmath76 , @xmath77 ) we show the 90% , 95% , 99% and 3@xmath78 cl contours for 2 d.o.f .
( @xmath79 and 11.83 , respectively )
. for @xmath80 and @xmath81 , the possibility of a sm interpretation along the @xmath82 axis ( taking into account systematic uncertainties ) would be excluded at greater than 99% cl for _ any _ assumed flux . in summary , we have found that observation of 1 earth - skimming and 10 quasi - horizontal events would exclude the ( perturbative ) sm at the 99% cl .
thus the expected low neutrino `` luminosity '' is not at all a show - stopper , and the observatory has the potential to uncover new physics at scales exceeding those accessible to the lhc .
if new non - perturbative physics exists , a decade or so would be required to uncover it at auger in the best case scenario ( cosmic neutrino flux at the waxman - bahcall level and @xmath24 cross section about an order of magnitude above the sm prediction ) .
of course future cr experiments should benefit from a much larger aperture , making such discovery conceivable in a short time scale .
an optimist may even imagine the possibility of probing qcd dynamics @xcite .
i would like to thank felix aharonian , markus ahlers , segev benzvi , hans blmer , mandy cooper sarkar , olivier deligny , tere dova , jonathan feng , haim goldberg , concha gonzalez garcia , yann guardincerri , francis halzen , nicole krohm , fred kuehn , antoine letessier selvon , jim matthews , teresa montaruli , giulia pancheri , tom paul , subir sarkar , viviana scherini , al shapere , paul sommers , john swain , diego torres , alan watson , and tom weiler for valuable discussions and permission to reproduce some of the figures .
i am grateful to stefanie pinnow for her proofreading skills and to my sister gaby for her assistance with slide management .
is supported by the u.s
. national science foundation ( nsf grant no phy-0757598 ) and the uwm research growth initiative . any opinions , findings , and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the national science foundation .
if uhecrs are heavy nuclei , the relic photons can excite the gdr at nucleus energies @xmath85 gev ( corresponding to @xmath86 in the nuclear rest frame ) , and thus there should be accompanying photo - dissociated free neutrons , themselves a source of @xmath87-decay antineutrinos .
the decay mean free path of a neutron is @xmath88 gev ) mpc , the lifetime being boosted from its rest frame value @xmath89 s to its lab value via @xmath90 . compared to cosmic distances @xmath91 mpc ,
the decay of even the boosted neutron may be taken as nearly instantaneous , and thus all free neutrons are themselves a source of @xmath87-decay cosmogenic antineutrinos .
the neutron emissivity @xmath92 defined as the mean number of neutrons emitted per comoving volume per unit time per unit energy as measured at the source can be estimated as follows .
neutrons with energies above @xmath93 gev have parent iron nuclei with @xmath94 which are almost completely disintegrated in distances of less than 100 mpc ( see sec . [ nuclei ] ) .
thus , it is reasonable to define a characteristic time @xmath95 given by the moment at which the number of nucleons is reduced to @xmath96 of its initial value @xmath97 , and presume the nucleus , emitted at distance @xmath98 from the earth , is a traveling source that at @xmath99 disintegrates into @xmath97 nucleons all at once @xcite .
then , the number of neutrons with energy @xmath100 can be approximated by the product of 1/2 the number of nucleons generated per nucleus and the number of nuclei emitted , _
i.e. _ @xmath101n@xmath102 , where @xmath103 is the mean neutron number of the source nucleus .
now , to obtain an estimate of the diffuse antineutrino flux one needs to integrate over the population of nucleus - emitting - sources out to the horizon @xcite @xmath104 \,\ , \int_0^q d\epsilon_{\overline \nu}\,\frac{dp}{d\epsilon_{\overline \nu } } ( \epsilon_{\overline \nu } ) \nonumber \\ & & \times \int_{-1}^1 \frac{d\cos \overline \theta_{\overline\nu}}{2 } \ ; \delta\left[e_{\overline \nu}-e_n\,\epsilon_{\overline \nu}\,(1+\cos \overline \theta_{\overline \nu } ) /m_n\right ] \ , , \label{nuflux}\end{aligned}\ ] ] where the @xmath105 in the volume integral is compensated by the usual @xmath106 fall - off of flux per source . here , @xmath107 is the hubble constant , @xmath108 and @xmath109 are the antineutrino and neutron energies in the lab , @xmath110 is the antineutrino angle with respect to the direction of the neutron momentum in the neutron rest - frame , and @xmath111 is the antineutrino energy in the neutron rest - frame . the last three variables are not observed by a laboratory neutrino - detector , and so are integrated over
. the observable @xmath112 is held fixed .
the delta - function relates the neutrino energy in the lab to the three integration variables .
the parameters appearing in eq .
( [ nuflux ] ) are the neutron mass and rest - frame lifetime ( @xmath113 and @xmath114 ) .
finally , @xmath115 is the normalized probability that the decaying neutron produces a @xmath116 with energy @xmath117 in the neutron rest - frame .
note that the maximum @xmath116 energy in the neutron rest frame is very nearly @xmath118 mev .
integration of eq .
( [ nuflux ] ) can be easily accomplished , especially when two good approximations are applied @xcite .
the first approximation is to think of the @xmath87decay as a @xmath119 process of @xmath120 in which the neutrino is produced monoenergetically in the rest frame , with @xmath121 mev , where @xmath122 mev is the neutron - proton mass difference . in the lab ,
the ratio of the maximum @xmath116 energy to the neutron energy is @xmath123 and so the boosted @xmath116 has a spectrum with @xmath124 the second approximation is to replace the neutron decay probability @xmath125 with a step function @xmath126 at some energy @xmath127 gev . combining these two approximations we obtain @xmath128 where @xmath129 and @xmath130 is the energy cutoff at the nucleus - emitting - source @xmath131 gev . for @xmath132 integration of eq .
( [ s1 ] ) leads to @xmath133 \ ,
\left . { \cal q}_a \right|_{\gamma_0}\,\ , , \label{znz}\ ] ] where @xmath134 gev and @xmath135 the sub - pev antineutrino spectrum is flat , as all the free neutrons have sufficient energy @xmath136 to contribute equally to all the @xmath116 energy bins below @xmath137 gev .
taking @xmath138 as a reasonable example , and inputting the observational value for iron nuclei , @xmath139 ev m@xmath140 s@xmath141 sr@xmath141 , eq .
( [ znz ] ) becomes @xmath142 note that the @xmath87-decay process gives initial antineutrino flavor ratios @xmath143 and earthly ratios nearly @xmath144 compared to full - blown monte carlo simulations @xcite , this back of the envelope calculation underestimates the flux by about 30% .
this is because the preceeding calculation does not account for possible neutrons produced in @xmath145 collisions .
of course the situation described above represents the most extreme case , in which all cosmic rays at the end of the spectrum are heavy nuclei . a more realistic guess
would be that the composition at the end of the spectrum is mixed .
photodisintegration of high energy nuclei is followed by immediate photo - emission from the excited daughter nuclei .
for brevity , we label the photonuclear process @xmath146 , followed by @xmath147-ray as `` @xmath148 '' @xcite .
the gdr decays dominantly by the statistical emission of a single nucleon , leaving an excited daughter nucleus @xmath149 .
the excited daughter nuclei typically de - excite by emitting one or more photons of energies @xmath150 mev in the nuclear rest frame .
the lab - frame energy of the @xmath151-ray is then @xmath152 , where @xmath153 is the boost factor of the nucleus in the lab . of interest here is the @xmath151-ray flux produced when
the photo - dissociated nuclear fragments produced on the cmb and cib de - excite .
these @xmath151-rays create chains of electromagnetic cascades on the cmb and cib , resulting in a transfer of the initial energy into the so - called fermi - lat region , which is bounded by observation to not exceed @xmath154 @xcite .
fortunately , we can finesse the details of the calculation by arguing in analogy to work already done .
the photodisintegration chain produces one @xmath87-decay neutrino with energy of order 0.5 mev in the nuclear rest frame , for each neutron produced . multiplying this result by 2 to include photodisintegration to protons in addition to neutrons
correctly weights the number of steps in the chain .
each step produces on average one photon with energy @xmath155 mev in the nuclear rest frame . in comparison ,
about 12 times more energy is deposited into photons . including the factor of 12 relating @xmath156 to @xmath157
, we find from ( [ buengusto ] ) that cosmogenic photodisintegration / de - excitation energy , @xmath158 ev/@xmath159 , is more than three orders of magnitude below the fermi - lat bound .
this result appears to be nearly invariant with respect to varying the maximum energy of the fe injection spectrum ( with a larger @xmath160 , the additional energy goes into cosmogenic pion production ) .
thus , there is no constraint on a heavy nuclei cosmic - ray flux from the @xmath148 mechanism .
neutron emission from a cosmological distribution of cr sources would also lead to a flux of @xmath161 .
in analogy to our previous estimate , we sum over the neutron - emitting sources out to the edge of the universe ; eq .
( [ s1 ] ) becomes @xcite @xmath162 where @xmath163 is the neutron volume emissivity .
an upper limit can be placed on @xmath164 by assuming an extreme situation in which all the crs escaping the source are neutrons , _ i.e. _ , @xmath165 } = \int de_n\,\ , e_n\,\ , { \cal q}_n ( e_n ) .$ ] with the production rate of ultrahigh energy protons ( [ professionally ] ) , and an assumed injection spectrum @xmath166 eq .
( [ ndk ] ) gives @xmath167 which is about three orders of magnitude below the waxman - bahcall bound .
em interactions of photons and leptons with the extragalactic radiation backgrounds and magnetic field can happen on time - scales much shorter than their production rates .
the relevant processes with background photons contributing to the differential interaction rates @xmath168 , @xmath169 , and @xmath170 are inverse compton scattering ( ics ) , @xmath171 , pair production ( pp ) , @xmath172 , double pair production ( dpp ) @xmath173 , and triple pair production ( tpp ) , @xmath174 @xcite
. high energetic electrons and positrons may also lose energy via synchrotron radiation in the intergalactic magnetic field @xmath175 with a random orientation @xmath176 with respect to the velocity vector .
we will assume in the following that the field strength @xmath177
g. , tpp by electrons can be neglected @xcite . for @xmath178 , dpp of photons
can also be safely neglect in the calculation @xcite .
] this leads to an efficient transfer of energy into the em cascade .
the synchrotron power spectrum ( w ev@xmath179 ) has the form @xmath180 where @xmath181 is the modified bessel function and @xmath182 .
this can be treated as a continuous energy loss of the electrons and positrons with a parameter = 0 $ ] , implying overall energy conservation . ]
@xmath183 we will assume in the calculation that the intergalactic magnetic field is primordial with a ( flux - conserving ) redshift dependence @xmath184 . note that the synchrotron energy loss has then a redshift dependence similar to bh pair production in the cmb , _
i.e. _ @xmath185 .
it is also convenient to define @xmath186 , which has an analogous redshift dependence , _ i.e. _ @xmath187
. the fast evolution of the cascade is governed by the set of differential equations , @xmath188 \ , , \nonumber \\
\partial_{\hat{t}}y_\gamma(e ) = & -\gamma_\gamma(e)y_\gamma(e ) + \int de'\frac{\mathcal{p}(e',e)}{e}y_e(e')+\int d e'\gamma_{e\gamma}(e',e)y_e(e ' ) \,,\label{eq : cas1}\end{aligned}\ ] ] which determines the evolution on _ short _ time - scales @xmath189 ( the redshift @xmath43 is kept _
fixed _ meanwhile ) .
the initial condition @xmath190 is given by the sum of previously developed cascades and the newly generated contributions from proton interactions .
the solution of the system ( [ eq : cas1 ] ) for an infinitesimally small step @xmath191 can be written for a discrete energy spectrum , @xmath192 , as @xmath193 with the transition matrix @xmath194 , defined by eq .
( [ eq : discrete ] ) , we can efficiently follow the development of the em cascade over a distance @xmath195 via matrix doubling @xcite : @xmath196^{2n}\,.\ ] ] the total energy of the cascade can be obtained by integrating eq .
( [ newdiff ] ) : @xmath197 = - \int d e e\partial_{\mathscr{e}}\left[b_{\rm cas}(z,\mathscr{e } ) \mathscr{z}_{\rm p}(z , e)\right]\,.\ ] ] integrating the r.h.s . by parts yields @xmath198 + \int d e \frac{1}{1+z}b_{\rm cas}(z,\mathscr{e } )
\mathscr{z}_p ( z , e)\,.\ ] ] the first term vanishes since @xmath199 for sufficiently low energies and @xmath200 beyond the maximal energy .
the time integration of the l.h.s . between the present epoch ( @xmath201 ) and
the first sources ( @xmath202 ) gives @xmath203 = \int d e e n_{\rm cas}(e ) = \omega_{\rm cas}\ , , \ ] ] hence we obtain eq .
( [ eq : omegacas ] ) . and @xmath204 ) , power index @xmath205 and evolution with @xmath206 and @xmath207 .
the fluxes are computed for 3 extragalactic magnetic field strengths : 1 pg , 1 ng , and 50 ng .
proton diffusion in extragalactic magnetic fields @xmath208 has not yet been included in the calculation .
we have neglected radiative losses of muons and pions before decaying into neutrinos ; the energy loss in muons is down by a factor @xmath209 compared to electrons .
this figure is courtesy of markus ahlers . ]
synchrotron radiation in strong magnetic fields , @xmath210 , can also by - pass the em cascade and transfer energy to sub - gev photons that are unconstrained by the fermi - lat spectrum @xcite .
this would be relevant for electrons around @xmath211 gev , where synchrotron loss starts to dominate over ics loss in the cmb ; see fig .
[ fig : bsamples ] . from eqs .
( [ synch ] ) and ( [ ginzburg ] ) we can verify that @xmath212 . of course , reaching a 50 ng field requires some astrophysical generator to augment the primordial field .
mpc , the current upper limit on primordial seed fields is about 1 ng @xcite . ] for simplicity , we have taken the redshift evolution of the generator to be the same as the evolution of the primordial field .
if were to assume an evolution which follows sfr , this will not significantly alter our conclusions .
the totem experiment is dedicated to the measurement of the total @xmath213 cross section with the luminosity independent method based on the optical theorem , @xmath214 where @xmath215 satisfies @xmath216 with @xmath217 the angle of the scattered proton with respect to the beam direction .
squaring eq .
( [ 0 ] ) we obtain @xmath218_{t=0}}{{\cal l}}\,,\ ] ] where @xmath219 is the integrated luminosity .
now , following @xcite , we can obtain the total cross section independently from @xmath219 , by using @xmath220 @xmath221_{t = 0 } \over ( n_{\rm el } + n_{\rm inel } ) } \,.\ ] ] here , @xmath222 and @xmath223 are the numbers of elastic and inelastic events , and @xmath224 is the ratio between the real and imaginary parts of the forward scattering amplitude @xcite . is an extrapolation to @xmath225 tev , and may be measured by the lhc experiments .
otherwise , it will contribute to the uncertainty in @xmath226 . ]
the difficult aspect of this measurement is obtaining a good extrapolation of the cross section for low momentum transfer . recall that @xmath227 , and so @xmath228 implies a measurement in the extreme forward direction .
the totem experiment aims to measure down to values of @xmath229 which corresponds to @xmath230rad @xcite .
in 1976 t hooft observed that the standard model does not strictly conserve baryon and lepton number @xcite .
rather , non - trivial fluctuations in @xmath231 gauge fields generate an energy barrier interpolating between topologically distinct vacua .
an index theorem describing the fermion level crossings in the presence of these fluctuations reveals that neither baryon nor lepton number is conserved during the transition , but only the combination @xmath232 inclusion of the higgs field in the calculation modifies the original instanton configuration @xcite .
an important aspect of this modification ( called the `` sphaleron '' ) is that it provides an explicit energy scale of about 10 tev for the height of the barrier .
this barrier can be overcome through thermal transitions at high temperatures @xcite , providing an important input to any calculation of cosmological baryogenesis .
more speculatively , it has been suggested @xcite that the topological transition could take place in two particle collisions at very high energy . the anomalous electroweak contribution to the partonic process
can be written as @xmath233 where @xmath234 , the tunneling suppression exponent @xmath235 is sometimes called the `` holy - grail function '' , and @xmath236 tev . thus , it is even possible that at or above the sphaleron energy the cross - section could be of @xmath237 @xcite .
of particular interest here would be an enhancement of the neutrino cross section over the perturbative sm estimates , say by an order of magnitude in the energy range @xmath238 .
the argument for strong damping of the anomalous cross section for @xmath239 30 tev was convincingly demonstrated in @xcite , in the case that the classical field providing the saddle point interpolation between initial and final scattering states is dominated by spherically symmetric configurations .
this @xmath240 symmetry allows the non - vacuum boundary conditions to be fully included in extremizing the effective action . in
@xcite it was shown that a _ sufficient _ condition for the @xmath240 dominance is that the interpolating field takes the form of a chain of `` lumps '' which are well - separated , so that the each lump lies well into the exponentially damped region of its nearest neighbors . however , we are not aware of any reason that such lumped interpolating fields should dominate the effective action .
it is thus of interest to explore the other extreme , in which non - spherically symmetric contributions dominate the effective action ( and let experiment be the arbiter ) .
it was shown @xcite that for the simple sphaleron configuration @xmath241-wave unitarity is violated for @xmath242 tev . if for @xmath243 tev we saturate unitarity in each partial wave , then this yields a geometric parton cross section @xmath244 , where @xmath245 is some average size of the classical configuration . as a fiducial value we take the core size of the manton - klinkhamer sphaleron , @xmath246 cm . in this simplistic model , the @xmath24 cross section is @xmath247 where @xmath248 . in the region
@xmath249 the pdf for the up and down quarks is well approximated by @xmath250 , so the expression for the cross section becomes @xmath251 where the last factor of 2/2 takes into account the ( mostly ) 2 contributing quarks @xmath252 in this range of @xmath253 and the condition that only the left - handed ones contribute to the scattering .
this is about 80 times the sm cross section .
of course this calculation is very approximate and the cross section can easily be smaller by a factor of 10 ( _ e.g. _ if @xmath245 is 1/3 of the fiducial value used ) .
j. abraham _ et al . _
[ pierre auger collaboration ] , nucl .
instrum .
a * 613 * , 29 ( 2010 ) .
j. a. abraham _ et al .
_ [ pierre auger collaboration ] , nucl .
instrum .
a * 620 * , 227 ( 2010 ) [ arxiv:0907.4282 ] .
p. abreu _ et al .
_ [ pierre auger collaboration ] , astropart .
phys . * 34 * , 368 ( 2011 ) [ arxiv:1010.6162 [ astro-ph.he ] ] .
m. aglietta _ et al .
_ [ eas - top collaboration ] , astropart .
* 21 * , 583 ( 2004 ) .
t. antoni _ et al . _ [ kascade collaboration ] , astropart .
* 24 * , 1 ( 2005 ) [ arxiv : astro - ph/0505413 ] .
w. d. apel _ et al _ [ kascade collaboration ] , astropart .
phys . * 31 * , 86 ( 2009 ) [ arxiv:0812.0322 [ astro - ph ] ] .
j. bluemer , r. engel and j. r. hoerandel , prog .
phys . * 63 * , 293 ( 2009 ) [ arxiv:0904.0725 [ astro-ph.he ] ] .
r. u. abbasi _ et al . _ [ hires collaboration ] , phys .
* 100 * , 101101 ( 2008 ) [ arxiv : astro - ph/0703099 ] .
j. abraham _ et al .
_ [ pierre auger collaboration ] , phys . rev .
* 101 * , 061101 ( 2008 ) [ arxiv:0806.4302 ] .
g. e. romero , j. a. combi , l. a. anchordoqui and s. e. perez bergliaffa , astropart .
phys . * 5 * , 279 ( 1996 ) [ arxiv : gr - qc/9511031 ] .
m. honda , astrophys .
j. * 706 * , 1517 ( 2009 ) [ arxiv:0911.0921 [ astro-ph.he ] ] .
m. ahlers , l. a. anchordoqui and s. sarkar , phys .
d * 79 * , 083009 ( 2009 ) [ arxiv:0902.3993 ] .
h. yksel , m. d. kistler , j. f. beacom and a. m. hopkins , astrophys . j. * 683 * , l5 ( 2008 ) [ arxiv:0804.4008 [ astro - ph ] ] .
k. s. capelle , j. w. cronin , g. parente and e. zas , astropart .
* 8 * , 321 ( 1998 ) [ arxiv : astro - ph/9801313 ] .
x. bertou , p. billoir , o. deligny , c. lachaud and a. letessier - selvon , astropart .
* 17 * , 183 ( 2002 ) [ arxiv : astro - ph/0104452 ] .
j. l. feng , p.
fisher , f. wilczek and t. m. yu , phys . rev .
lett . * 88 * , 161102 ( 2002 ) [ arxiv : hep - ph/0105067 ] .
d. fargion , astrophys .
j. * 570 * , 909 ( 2002 ) [ arxiv : astro - ph/0002453 ] .
l. a. anchordoqui , j. l. feng , h. goldberg and a. d. shapere , phys .
d * 66 * , 103002 ( 2002 ) [ arxiv : hep - ph/0207139 ] .
r. gandhi , c. quigg , m. h. reno and i. sarcevic , phys .
d * 58 * , 093009 ( 1998 ) [ arxiv : hep - ph/9807264 ] .
g. pancheri and y. srivastava , phys .
b * 159 * , 69 ( 1985 ) .
g. pancheri and y. n. srivastava , phys .
b * 182 * , 199 ( 1986 ) . c. albajar
_ et al . _
[ ua1 collaboration ] , nucl .
b * 309 * , 405 ( 1988 ) . c. adloff
_ et al . _
[ h1 collaboration ] , eur .
j. c * 21 * , 33 ( 2001 ) [ arxiv : hep - ex/0012053 ] .
l. anchordoqui , m. t. dova , a. g. mariazzi , t. mccauley , t. c. paul , s. reucroft and j. swain , annals phys .
* 314 * , 145 ( 2004 ) [ arxiv : hep - ph/0407020 ] .
t. k. gaisser and t. stanev , phys .
b * 219 * , 375 ( 1989 ) .
r. s. fletcher , t. k. gaisser , p. lipari and t. stanev , phys .
d * 50 * , 5710 ( 1994 ) . n. n. kalmykov , s. s. ostapchenko and a. i. pavlov , nucl .
* 52b * , 17 ( 1997 ) .
k. eggert , nucl .
suppl . * 122 * , 447 ( 2003 ) .
atlas collaboration , _ measurement of the inelastic proton - proton cross section at @xmath254 tev with the atlas detector , _ atlas - conf-2011 - 002 ( 2011 ) .
m. ave , r. a. vazquez , e. zas , j. a. hinton and a. a. watson , astropart .
* 14 * , 109 ( 2000 ) [ arxiv : astro - ph/0003011 ] .
a. v. plyasheshnikov and f. a. aharonian , j. phys .
g * 28 * , 267 ( 2002 ) [ arxiv : astro - ph/0107592 ] . a. kusenko and t. j. weiler , phys .
lett . * 88 * , 161101 ( 2002 ) [ arxiv : hep - ph/0106071 ] . l. a. anchordoqui , j. l. feng , h. goldberg and a. d. shapere , phys .
d * 65 * , 124027 ( 2002 ) [ arxiv : hep - ph/0112247 ] .
d. gora , m. roth and a. tamburro , astropart .
* 26 * , 402 ( 2007 ) .
a. gazizov and m. p.
kowalski , comput .
commun . * 172 * , 203 ( 2005 ) [ arxiv : astro - ph/0406439 ] .
a. cooper - sarkar and s. sarkar , jhep * 0801 * , 075 ( 2008 ) [ arxiv:0710.5303 [ hep - ph ] ] .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * , 026 ( 2006 ) [ arxiv : hep - ph/0603175 ] .
s. jadach , z. was , r. decker and j. h. kuhn , comput .
commun . *
76 * , 361 ( 1993 ) .
consortium for spatial information ( cgiar - csi ) .
m. gluck , s. kretzer and e. reya , astropart .
phys . * 11 * , 327 ( 1999 ) [ arxiv : astro - ph/9809273 ] .
m. nadolsky _ et al .
_ , phys .
d * 78 * , 013004 ( 2008 ) [ arxiv:0802.0007 [ hep - ph ] ] .
n. armesto , c. merino , g. parente and e. zas , phys .
d * 77 * , 013001 ( 2008 ) [ arxiv:0709.4461 [ hep - ph ] ] .
t. j. weiler , nucl .
suppl . * 134 * , 47 ( 2004 ) . l. a. anchordoqui , z. fodor , s. d. katz , a. ringwald and h. tu , jcap * 0506 * , 013 ( 2005 ) [ arxiv : hep - ph/0410136 ] .
l. a. anchordoqui , h. goldberg , f. halzen and t. j. weiler , phys .
b * 593 * , 42 ( 2004 ) [ arxiv : astro - ph/0311002 ] .
d. hooper , a. taylor and s. sarkar , astropart .
* 23 * , 11 ( 2005 ) [ arxiv : astro - ph/0407618 ] .
m. ave , n. busca , a. v. olinto , a. a. watson and t. yamamoto , astropart .
phys . * 23 * , 19 ( 2005 ) [ arxiv : astro - ph/0409316 ] .
g. t hooft , phys .
lett . * 37 * , 8 ( 1976 ) . g. t hooft , phys .
d * 14 * , 3432 ( 1976 ) [ erratum - ibid .
d * 18 * , 2199 ( 1978 ) ] .
f. r. klinkhamer and n. s. manton , phys .
d * 30 * , 2212 ( 1984 ) .
v. a. kuzmin , v. a. rubakov and m. e. shaposhnikov , phys .
b * 155 * , 36 ( 1985 ) .
m. fukugita and t. yanagida , phys .
b * 174 * , 45 ( 1986 ) .
p. b. arnold and l. d. mclerran , phys .
d * 36 * , 581 ( 1987 ) .
h. aoyama and h. goldberg , phys .
b * 188 * , 506 ( 1987 ) .
a. ringwald , nucl .
b * 330 * , 1 ( 1990 ) .
o. espinosa , nucl .
b * 343 * , 310 ( 1990 ) .
a. ringwald , jhep * 0310 * , 008 ( 2003 ) [ arxiv : hep - ph/0307034 ] .
f. l. bezrukov , d. levkov , c. rebbi , v. a. rubakov and p. tinyakov , phys .
b * 574 * , 75 ( 2003 ) [ arxiv : hep - ph/0305300 ] .
t. m. gould and s. d. h. hsu , mod .
lett . a * 9 * , 1589 ( 1994 ) [ arxiv : hep - ph/9311291 ] . | this is a written version of a series of lectures aimed at graduate students in astrophysics / particle theory / particle experiment . in the first part
, we explain the important progress made in recent years towards understanding the experimental data on cosmic rays with energies @xmath0 .
we begin with a brief survey of the available data , including a description of the energy spectrum , mass composition , and arrival directions . at this point
we also give a short overview of experimental techniques .
after that , we introduce the fundamentals of acceleration and propagation in order to discuss the conjectured nearby cosmic ray sources , and emphasize some of the prospects for a new ( multi - particle ) astronomy .
next , we survey the state of the art regarding the ultrahigh energy cosmic neutrinos which should be produced in association with the observed cosmic rays . in the second part , we summarize the phenomenology of cosmic ray air showers .
we explain the hadronic interaction models used to extrapolate results from collider data to ultrahigh energies , and describe the prospects for insights into forward physics at the large hadron collider ( lhc ) .
we also explain the main electromagnetic processes that govern the longitudinal shower evolution .
armed with these two principal shower ingredients and motivation from the underlying physics , we describe the different methods proposed to distinguish primary species .
in the last part , we outline how ultrahigh energy cosmic ray interactions can be used to probe new physics beyond the electroweak scale .
= 1 |
pep - ii is an asymmetric b - factory that consists of two separate rings with different energies .
the electron and positron beams are brought into head - on collisions at the babar detector as shown in fig .
[ fig : pepii - ip ] . in order to separate the beams fast enough away from the interaction point ( ip ) to avoid the deteriorating effect on the luminosity due to adjacent parasitic collisions , the beams go through many strong magnets inside the solenoid with very large offsets from their center .
these offsets can be as large as a few centimeters as indicated in fig .
[ fig : pepii - ip ] .
the excursion of the design orbit introduces large uncertainty into the optics near the ip .
most problematic , the optics changes , as recently seen in the beam - beam experiments @xcite , when the local orbit varies .
an accurate optical model requires a map of three - dimensional magnetic field in the region . that is rather difficult to compute because of the overlapping fields and complicated geometry . as a first step
, we will use the hard - edge model for the fringe field to estimate the effects analytically in this paper .
the magnetic field of a single normal harmonics , including its fringe field , in the cylindrical coordinate is derived by bassetti and biscari @xcite . to study the single - particle effects of the fringe field using hamiltonian system , we need to know its corresponding vector potential . in the coulomb gauge , @xmath1 , the potential can be expressed as @xmath2 where @xmath3 for a skew multipole , the expression can be obtained by an exchange between sine and cosine . the coulomb gauge is chosen because its potential becomes the conventional multipole expansion . for the solenoid , due to its axial symmetry , it is more convenient to choose the axial gauge : @xmath5 .
the vector potential is given by @xmath6 the potential satisfies maxwell s equation @xmath7 .
any truncation of the series could violate maxwell s equation .
the magnetic field is given by @xmath8 .
in the cartesian coordinate system , the hamiltonian , using the distance @xmath9 as the independent variable for a charged particle moving in a static magnetic field , is given by @xcite @xmath10 where @xmath11 are scaled components of the vector potential along axis x , y , s , respectively ; @xmath12 are the transverse canonical momenta scaled by a reference momentum @xmath13 , @xmath14 , and @xmath15 is the path length .
we expand the square root in eq .
( [ eqn : h ] ) and keep only the first order of the vector potential , @xmath16 .
\label{eqn : hamiltonian}\end{aligned}\ ] ] this hamiltonian is used to compute the dynamical effects on the charged particles due to the fringe field .
taking a solenoid with field @xmath17 as an example and follow the method used by forest and milutinovic @xcite , we choose a hard - edge model @xmath18 where @xmath19 is the unit step function . using this model and the vector potential in eq .
( [ eqn : asolenoid ] ) , we have , @xmath20 and @xmath21 substituting these components into the hamiltonian in eq .
( [ eqn : hamiltonian ] ) , we obtain @xmath22 where @xmath23 and @xmath24 . after the standard manipulation of map and integration by parts @xcite
, we obtain the map @xmath25 : } = e^{:f_s : } , \nonumber\end{aligned}\ ] ] where @xmath26\ ] ] here , @xmath27 $ ] and @xmath28 $ ] denotes the poisson bracket . note that @xmath29 is invariant under the two - dimensional rotation around the axis of the solenoid . similarly , for a dipole magnet , we start with @xmath30 and set @xmath31 in eq .
( [ eqn : amultipole ] ) to obtain the components of the vector potential of a dipole magnet as follows @xmath32 where @xmath33 is the magnetic field of the dipole .
the hamiltonian is derived by substituting the vector potential into eq .
( [ eqn : hamiltonian ] ) .
we have @xmath34 where @xmath35 , \nonumber \\ v_2 & = & -{x\over8\rho}(x^2+y^2),\end{aligned}\ ] ] and @xmath36 and @xmath37 is the bending radius of the dipole magnet .
the final map is written as @xmath38 : } = e^{:f_d : } , \nonumber\ ] ] where , @xmath39.\ ] ] as a first - order kick and @xmath40 , it agrees with the expression found by papaphilippou , wei , and talman @xcite .
similar exercise can be carried out for a quadrupole magnet with @xmath41 .
the final map is given by @xmath42 where , @xmath43 .
\label{eqn : quadrupole}\ ] ] where @xmath44 and @xmath45 is the gradient .
@xmath46 agrees with the expression first found by forest and milutinovic . here
, the maps are derived for the raising edge of the magnet . for the falling edge ,
the maps are obtained by simply switching the sign of the lie generator of the map .
if there is a design off - axis orbit : @xmath47 and @xmath48 in the horizontal plane , the non - linear effects of the fringe field of a quadrupole magnet will feed down to linear optical errors . by substituting @xmath49 with @xmath50 and @xmath51 with @xmath52 into eq .
( [ eqn : quadrupole ] ) , and extracting the quadratic terms of @xmath53 , we find that the tune shifts are given by @xmath54 where @xmath55 and @xmath56 are the courant - snyder parameters at the position of the edge . for pep - ii ,
the estimated tune shifts in the vertical plane relative to the design orbit in the low energy ring are tabulated in table [ tab : tuneshift ] .
.tune shift from the quadrupole adjacent to the ip . [ cols="<,^,^,^,^",options="header " , ] [ tab : tuneshift ] note that the outside edges of the quadrupole contribute more because the excursions are larger .
these rather large optical effects are not currently included in our optical model .
the nonlinear fringe transformation at quadrupole edges has been recently implemented in the lego code @xcite .
based on eq .
( [ eqn : quadrupole ] ) , the fringe octupole - like field would generate an amplitude dependent tune shift and excite chromo - geometric octupole resonances .
these effects were observed in pep - ii dynamic aperture calculations . to maximize luminosity ,
the pep - ii horizontal tune is moved close to a half - integer .
however , the tune space in this region is limited by the effects of half - integer resonance and its synchrotron side bands .
the effect of the quadrupole resonances on pep - ii dynamic aperture had been observed in earlier tracking studies @xcite . in this case , the resonance condition is @xmath57 , where @xmath58 is the synchrotron tune . after including the nonlinear fringe in quadrupoles
, the tracking showed a reduction of dynamic aperture for off - momentum particles .
an example of dynamic aperture for pep - ii high energy ring ( her ) is shown in fig .
[ fig : aper ] . in this case , the @xmath59 her upgrade lattice @xcite is used where the ip beta functions and tunes are @xmath60 cm and @xmath61 .
the tracking included synchrotron oscillations , machine errors and various optics corrections . in fig .
[ fig : aper ] , the blue dotted lines show on - momentum dynamic aperture for 10 random error settings , the dash red lines correspond to relative momentum error of @xmath62 , and the solid ellipse shows the @xmath63 beam size for reference .
one can see that the fringe effect increases the on - momentum aperture but reduces the horizontal off - momentum aperture at .510 tune .
the on - momentum improvement is due to the fringe compensation of the amplitude dependent tune shift from sextupoles .
the off - momentum effect was attributed to the 1st octupole side band of the half - integer resonance excited by the fringe . in this case , the octupole resonance condition @xmath64 yields the resonance tune at @xmath65 . moving the tune away from this resonance to 28.518
restored the aperture above the @xmath63 size .
we have found a new lie generator for the hard - edge fringe due to a solenoid magnet . it is a fourth - order generator and gives an octupole - like kick to charge particles .
it can also provide additional x - y couplings through an off - centered orbit .
9 w. kozanecki _
_ , experimental study of crossing - angle and parasitic crossing effects at the pep - ii @xmath66 collider , " pac05 , these proceedings .
m. bassetti and c. biscari , pa * 52 * , 221 ( 1996 ) .
r. ruth , aip conference proceedings no .
153 , vol.1 p166 , m. month and m. dienes editors ( 1985 ) .
y. papaphilippou , j. wei , and r. talman , phys .
e * 67 * , 046502 ( 2003 ) .
e. forest and j. milutinovic , methods phys .
* a269 * , 474 ( 1988 ) . y. cai , _ et al .
_ , slac pub7642 ( 1997 ) .
y. cai , y. nosochkov , slac pub9812 ( 2003 ) .
y. cai , _ et al .
_ , slac pub9810 ( 2003 ) . | the leading lie generators , including the chromatic effects , due to hard - edge fringe field of single multipole and solenoid are derived from the vector potentials within a hamiltonian system .
these nonlinear generators are applied to the interaction region of pep - ii to analyze the linear errors due to the feed - down from the off - centered quadrupoles and solenoid .
the nonlinear effects of tune shifts at large amplitude , the synchro - betatron sidebands near half integer and their impacts on the dynamic aperture are studied in the paper . |
one of the most important and difficult problems in neuroscience is to understand how neurons communicate in a brain .
there has been a long - standing controversy between the temporal- and rate - code hypotheses in which information is assumed to be encoded in firing timings and rates , respectively @xcite-@xcite .
a recent success in brain - machine interface ( bmi ) @xcite , however , suggests that the population rate code is employed in sensory and motor neurons , though it is still controversial which of rate , temporal or other codes is adopted in higher - level cortical neurons . in recent years
, much attention has been paid to a study on effects of mean firing rate , fluctuation and synchrony ( or spatial correlation ) of input signals ( for a review on rate and synchrony , see ref .
the precise role of synchrony in information transmission and the relation among the firing rate , fluctuation and synchrony are not clear at the moment @xcite-@xcite .
the firing rate and synchrony are reported to be simultaneously modulated by different signals .
for example , in motor tasks of monkey , firing rate and synchrony are considered to encode behavioral events and cognitive events , respectively @xcite . during visual tasks , rate and synchrony
are suggested to encode task - related signals and expectation , respectively @xcite .
a change in synchrony may amplify behaviorally relevant signals in v4 of monkey @xcite .
an increase in synchrony of input signals is expected to yield an increase in output firing rate .
the synchrony of neurons in extrastriate visual cortex is , however , reported to be modulated by selective attention even when there is only small change in firing rate @xcite .
rate - independent modulations in synchrony are linked to expectation , attention and livalry @xcite .
fluctuations of input signals have been reported to modify the @xmath0 relation between an applied dc current @xmath1 and autonomous firing frequency @xmath2 although its sensitivity to input fluctuation seems to depend on a kind of neurons @xcite-@xcite .
the @xmath0 curve of prefrontal cortex ( pfc ) retains the increased sensitivity to input fluctuations at large @xmath1 , while that of somatosensory cortex ( ssc ) is insensitive to input fluctuation though its linearity is increased at small @xmath1 @xcite .
this kind of problems discussed above have been extensively studied by using spiking neuron models such as the hodgkin - huxley model @xcite and integrate - and - fire ( if ) model with diffusion and mean - field approximations @xcite-@xcite .
the purpose of the present paper is to examine the same problem by using the rate - code model , which is an alternative theoretical model to the spiking model . in a previous paper
@xcite ( referred to as i hereafter ) , we proposed the generalized rate - code model for coupled neuron ensembles with finite populations , which are subjected to additive and multiplicative noises .
it seems natural to include multiplicative noise beside additive noise in our rate - code model because the noise intensity is expected to generally depend on the state of neurons .
actually effects of multiplicative noise in the spiking neuron model are extensively examined by using conductance - based inputs which yield multiplicative noise @xcite-@xcite .
our calculation in i has shown that the introduced multiplicative noise leads to the non - gaussian stationary distribution of firing rate , yielding interspike - interval ( isi ) distributions such as the gamma , inverse - gaussian and log - normal distributions , which have been experimentally observed .
we discussed in i the dynamical properties of neuron population , by using the augmented moment method ( amm ) which was developed for a study of stochastic systems with finite populations @xcite . in the amm , we pay attention to global properties of neuronal ensembles , taking account of mean and fluctuations of local and global variables .
the amm has the same purpose to effectively study the properties of neuronal ensembles as approaches based on the population - code hypothesis @xcite-@xcite .
the amm has been nicely applied to various subjects of neuronal ensembles @xcite and complex networks @xcite .
we have assumed in i that input signals are the same for all neurons in the ensemble . in the present study ,
input signals are allowed to fluctuate and to be spatially correlated .
we will derive equations of motion for mean , fluctuation and synchrony of firing rates with the use of the amm , in order to investigate the response to mean- , fluctuation- and synchrony - driven inputs .
this study clarifies , to some extent , their respective roles in information transmission .
the paper is organized as follows . in sec .
2 , we discuss an application of the amm to the generalized rate model , studying the input - output relations of mean .
fluctuation and synchrony . in sec . 3 ,
stationary and dynamical properties are discussed with numerical model calculations . by using the independent component analysis ( ica ) @xcite
, we investigate a separation of signals when two or three kinds of inputs are simultaneously applied .
discussions are presented in sec .
4 , where the results obtained by spiking if model are compared with those by our rate - code model . the final sec . 5
is devoted to conclusion .
for a study of the properties of a neuron ensemble containing finite @xmath3 neurons , we have adopted the generalized rate - code model @xcite in which a neuron is regarded as a transducer from input to output signals , both of which are expressed in terms of firing rates .
the dynamics of the firing rate @xmath4 ( @xmath5 ) of a neuron @xmath6 ( @xmath7 to @xmath3 ) is given by @xmath8 with @xmath9 here @xmath10 and @xmath11 are arbitrary functions of @xmath12 , @xmath13 @xmath14 denotes the coordination number , @xmath15 an input signal from external sources , @xmath16 the coupling strength , and @xmath17 the heaviside function : @xmath18 for @xmath19 and @xmath20 otherwise .
additive and multiplicative noises are included by @xmath21 and @xmath22 , respectively , expressing zero - mean gaussian white noise with correlations given by @xmath23\delta(t - t ' ) , \label{eq : a4 } \\
\left < \xi_{i}(t)\:\xi_{j}(t ' ) \right > & = & \beta^2 [ \delta_{ij}+c_0(1-\delta_{ij } ) ] \delta(t - t ' ) , \label{eq : a5 } \\
\left < \eta_{i}(t)\:\xi_{j}(t ' ) \right > & = & 0,\end{aligned}\ ] ] where the bracket @xmath24 denotes the average over the distribution of @xmath25 [ eq . ( [ eq : x1 ] ) ] , @xmath26 ( @xmath27 ) stands for the magnitudes of multiplicative ( additive ) noise , and @xmath28 ( @xmath29 ) expresses the degree of the spatial correlation in multiplicative ( additive ) noise .
the gain function @xmath30 in eq .
( [ eq : a3 ] ) expresses the response of a rate output ( @xmath12 ) to a rate input ( @xmath31 ) .
it has been shown that , when spike inputs with mean firing rate @xmath32 are applied to a hodgkin - huxley neuron , mean firing rate @xmath33 of output signals is @xmath34 for @xmath35 [ hz ] , above which @xmath33 shows the saturation behavior @xcite .
the nonlinear , saturating behavior in @xmath30 arises from the fact that a neuron can not fire with the rate of @xmath36 @xmath37 where @xmath38 denotes the refractory period . the function @xmath30 has the rectifying property because the firing rate is positive , which is expressed by the heaviside function in eq .
( [ eq : a3 ] ) .
although our results to be present in the following are valid for any choice of @xmath39 , we have adopted , in this study , a simple analytic expression given by eq .
( [ eq : a3 ] ) , where the rate is normalized by @xmath40 . with the use of the diffusion - type approximation , a spatially correlated input signal @xmath15 in eq .
( [ eq : a2 ] ) is assumed to be given by @xmath41 with @xmath42 \delta(t - t ' ) , \label{eq : a8}\end{aligned}\ ] ] where variance ( @xmath43 ) and covariance ( @xmath44 ) are given by @xmath45 @xmath46 expressing the degree of the spatial correlation .
we will discuss responses of the neuronal ensemble described by eqs .
( [ eq : a1])-([eq : a3 ] ) to the spatially correlated input given by eqs .
( [ eq : a6])-([eq : a8 ] ) with given @xmath47 , @xmath43 and @xmath48 , by using both direct simulations ( dss ) and the amm @xcite . in the amm @xcite
, we define the three quantities of @xmath49 , @xmath50 and @xmath51 given by @xmath52 ^ 2 \rangle , \label{eq : b2}\\
\rho(t ) & = & \frac{1}{n^2 } \sum_i \sum_j \langle [ r_i(t)-\mu(t)][r_j(t)-\mu(t ) ] \rangle , \label{eq : b3 } \end{aligned}\ ] ] where @xmath53 , @xmath54 expresses the mean , and @xmath55 and @xmath56 denote the averaged , auto and mutual correlations , respectively , of firing rates . by using the fokker - planck equation ( fpe ) , we obtain equations of motion for @xmath57 and @xmath58 ( for details see appendix a ) . expanding @xmath32 in eqs .
( [ eq : x2 ] ) and ( [ eq : x3 ] ) around the average value of @xmath49 as @xmath59 and retaining up to the order of @xmath60 , we obtain equations of motion for @xmath49 , @xmath50 and @xmath51 .
amm equations in the stratonovich representation are given by ( the argument @xmath61 being suppressed)@xmath62 , \label{eq : b6}\\ \frac{d \gamma}{dt } & = & 2f_{1 } \gamma + \frac{2 h_{1 } w}{z } \left(n \rho-\gamma
\right ) + 2(g_{1}^2 + 2 g_{0}g_{2})\alpha^2\gamma + \gamma_i+ \alpha^2 g_{0}^2+\beta^2 , \label{eq : b7}\\ \frac{d \rho}{dt } & = & 2 f_{1 } \rho + 2 h_1 w \rho + 2 ( g_{1}^2 + 2 g_{0}g_{2 } ) \alpha^2 \rho \nonumber \\ & + & \frac{1}{n}(\gamma_i+\alpha^2 g_0 ^ 2+\beta^2 ) + \frac{z}{n}(\gamma_i s_i+c_1 \alpha^2 g_0 ^ 2 + c_0 \beta^2 ) , \label{eq : b8 } \ ] ] where @xmath63 , @xmath64 , @xmath65 and @xmath66 .
original @xmath3-dimensional stochastic des given by eqs .
( [ eq : a1])-([eq : a3 ] ) are transformed to the three - dimensional deterministic des given by eqs .
( [ eq : b6])-([eq : b8 ] ) . for @xmath67 ,
equations of motion given by eqs .
( [ eq : b6])-([eq : b8 ] ) reduce to those obtained in our previous study @xcite . from @xmath49 , @xmath50 and @xmath51 obtained by eqs .
( [ eq : b6])-([eq : b8 ] ) , we may calculate important quantities of synchrony and variability . then
, they may be expressed in physically more transparent forms , as will be discussed in the following [ eqs .
( [ eq : d1])-([eq : d3 ] ) ] .
the synchronization is conventionally discussed for firing timings ( _ temporal synchronization _ ) or phase ( _ phase synchronization _ ) .
we discuss , in this paper , the synchronization for firing rate ( _ rate synchronization _ ) . in order to quantitatively discuss the synchronization , we first consider the quantity @xmath68 given by @xmath69 ^ 2 > = 2 [ \gamma(t)-\rho(t ) ] .
\label{eq : c1}\ ] ] when all neurons are firing with the same rate ( the completely synchronous state ) , we obtain @xmath70 for all @xmath6 , and then @xmath71 in eq .
( [ eq : c1 ] ) . on the contrary , we obtain @xmath72 in the asynchronous state where @xmath73 @xcite .
we may define the normalized ratio for the synchrony of firing rates given by @xcite @xmath74 @xmath75 is 0 and 1 for completely asynchronous ( @xmath76 ) and synchronous states ( @xmath77 ) , respectively .
the variability in the isi is usually defined by by @xmath78 where @xmath79 and @xmath80 stand for mean and variance of isi , respectively .
we here define the rate variability given by @xmath81 similarly , the variability in input rate signals is given by @xmath82 we may show that if the temporal firing rate is given by @xmath83 , we obtain @xmath84 , where @xmath85 denotes the isi of firing times in a neuron @xmath6 . it is noted that the variability and synchrony are related to the second - order statistics of local ( @xmath50 ) and global fluctuations ( @xmath51 ) , respectively , of firing rates . employing the relations given by eq .
( [ eq : c2 ] ) , we may transform equations of motion for @xmath49 , @xmath50 and @xmath51 given by eqs .
( [ eq : b6])-([eq : b8 ] ) to those for @xmath49 , @xmath50 and @xmath86 given by @xmath87 , \label{eq : d1}\\ \frac{d \gamma}{dt } & = & 2f_{1 } \gamma + 2h_{1 } w \gamma s + 2(g_{1}^2 + 2 g_{0}g_{2})\alpha^2\gamma + \gamma_i+ \alpha^2 g_{0}^2 + \beta^2 , \label{eq : d2}\\ \frac{d s}{dt } & = & - \frac{1}{\gamma}(\gamma_i+ \alpha^2 g_0 ^ 2+\beta^2)s + \left ( \frac{1}{\gamma } \right ) ( \gamma_i s_i+c_1 \alpha^2 g_0 ^ 2+c_0 \beta^2 ) \nonumber \\ & + & \left ( \frac{2 h_1 w}{z } \right ) ( 1+z s)(1-s ) .
\label{eq : d3 } \end{aligned}\ ] ] equations ( [ eq : d1])-([eq : d3 ] ) express a response of @xmath88 to a given input of @xmath89 .
input signals in which information is encoded in @xmath90 , @xmath43 and @xmath91 are hereafter referred to as _ mean - driven _ , _ fluctuation - driven _ and _ synchrony - driven _ inputs , respectively
. equations ( [ eq : d1])-([eq : d3 ] ) show that @xmath43 plays the same role as independent noise while @xmath92 as correlated noise . when we adopt @xmath10 and @xmath11 given by @xmath93 eqs .
( [ eq : d1])-([eq : d3 ] ) become @xmath94 , \label{eq : d6}\\ \frac{d \gamma}{dt } & = & -2 \lambda a \mu^{a-1 } \gamma + 2 h_1 w \gamma s + 2 b ( 2b-1 ) \alpha^2 \mu^{2b-2}\gamma + \gamma_i + \alpha^2 \mu^{2b } + \beta^2 , \label{eq : d7 } \\
\frac{d s}{dt } & = & - \frac{1}{\gamma } ( \gamma_i + \alpha^2 \mu^{2b}+\beta^2)s + \left ( \frac{1}{\gamma } \right ) ( \gamma_i s_i+c_1 \alpha^2 \mu^{2b}+c_0 \beta^{2 } ) \nonumber \\ & + & \left ( \frac { 2 h_1 w}{z } \right ) ( 1+zs)(1-s ) , \label{eq : d8}\end{aligned}\ ] ] where @xmath95 expresses the relaxation rate .
we note that for ( i ) @xmath96 or @xmath97 and ( ii ) @xmath98 , @xmath99 or @xmath97 , a motion of @xmath49 is decoupled from the rest of variables because the @xmath100 and @xmath101 in the third and fourth terms of eq .
( [ eq : d6 ] ) vanish .
equation ( [ eq : d8 ] ) shows that a motion of @xmath86 is ostensibly independent of the index @xmath102 of @xmath10 although it depends on @xmath102 through @xmath50 .
in order to get an insight to the present method , we will show some model calculations . when we consider a special case of @xmath103 in eqs .
( [ eq : d4 ] ) and ( [ eq : d5 ] ) , @xmath104 eqs .
( [ eq : d6])-([eq : d8 ] ) are expressed by @xmath105 the stationary solution of eqs .
( [ eq : g3])-([eq : g5 ] ) is given by @xmath106+h_1 w \gamma_i s_i } , \label{eq : g8 } \\ & = & \frac{(\gamma_i s_i+ c_1 \alpha^2 \mu^2+c_0 \beta^2 ) } { ( \gamma_i+\alpha^2 \mu^2+\beta^2 ) } , \hspace{1cm}\mbox{for $ w=0$}\\ & = & \frac{h_1 w}{[z(\lambda-\alpha^2)-h_1 w ( z-1)]}. \hspace{0.5cm}\mbox{for $ c_0=c_1=s_i=0$}\end{aligned}\ ] ] we note in eq .
( [ eq : g6 ] ) that @xmath49 is increased as @xmath47 is increased with an enhancement factor of @xmath107 , but @xmath49 is independent of @xmath43 and @xmath48 .
local fluctuation @xmath50 is increased with increasing input fluctuation ( @xmath43 ) and/or noise ( @xmath26 , @xmath27 ) as eq .
( [ eq : g7 ] ) shows . equation ( [ eq : g8 ] )
shows that @xmath86 is increased with increasing @xmath48 , @xmath29 , @xmath28 and/or @xmath16 , as expected .
the stability condition around the stationary state may be examined from eigenvalues of the jacobian matrix of eqs .
( [ eq : g3])-([eq : g5 ] ) , which are given by ( for details , see appendix b ) @xmath108 the first eigenvalue of @xmath109 arises from an equation of motion for @xmath49 , which is decoupled from the rest of variables . the stability condition for @xmath49
is given by @xmath110 in contrast , the stability condition for @xmath50 and @xmath51 is given by @xmath111 then for @xmath112 , @xmath50 and @xmath51 are unstable but @xmath49 remains stable .
the parameters in our model are @xmath26 , @xmath27 , @xmath16 and @xmath3 : we hereafter set @xmath113 and @xmath114 to reduce the number of model parameters .
input signals are characterized by @xmath47 , @xmath43 and @xmath48 .
we will present some numerical calculations for mean- , fluctuation- and synchrony - driven inputs which are calculated with the use of eqs .
( [ eq : g6])-([eq : g8 ] ) for @xmath115 , @xmath116 and @xmath117 . * a. mean - driven inputs * figure [ figa ] shows the @xmath47 dependences of @xmath49 and @xmath86 for @xmath118 ( dashed curves ) and @xmath119 ( solid curves ) with @xmath120 and @xmath121 .
we note that for @xmath122 , @xmath49 is increased with increasing @xmath47 after the gain function of @xmath30 , while @xmath86 is independent of @xmath47 .
in contrast , for finite @xmath16 , @xmath49 is much increased than that for @xmath122 .
the chain line expressing @xmath123 crosses the @xmath49 curve at @xmath124 for @xmath118 and at @xmath125 for @xmath119 , * b. fluctuation - driven inputs * figure [ figb ] shows @xmath126 and @xmath86 against @xmath127 ( @xmath128 ) for @xmath118 ( dashed curves ) and @xmath119 ( solid curves ) with @xmath129 and @xmath121 . for @xmath118 , an increase in @xmath127
yields an increase in @xmath126 and @xmath86 , while @xmath49 is independent of @xmath130 , as eqs .
( [ eq : g6])-([eq : g8 ] ) show .
the chain curve shows @xmath131 : the region where @xmath132 is realized for @xmath133 for @xmath118 and @xmath134 for @xmath119 . *
c. synchrony - driven inputs * figure [ figc ] shows the @xmath48 dependences of @xmath49 and @xmath86 for @xmath118 ( dashed curves ) and @xmath119 ( solid curves ) with @xmath129 and @xmath120 . with increasing @xmath48 , @xmath86 is increased as expected . for @xmath122 ,
@xmath49 is independent of @xmath48 , as eqs .
( [ eq : g6 ] ) shows . for @xmath119 , @xmath86
is much increased compared to that for @xmath118 .
it is necessary to point out that the @xmath47 dependence of @xmath49 is modified by multiplicative noise ( @xmath26 ) .
an example of the @xmath47 dependence of @xmath49 is plotted in fig .
[ figd ] for various @xmath26 . with increasing @xmath26
, @xmath49 shows a steeper increase for larger @xmath26 because of the @xmath135 factor in eq .
( [ eq : g6 ] ) .
this reminds us the recent experiment of prefrontal cortex ( pfc ) showing that the @xmath0 curve has the increased sensitivity at large @xmath1 with increasing input fluctuation @xcite .
this is interpreted as due to a shorten , effective refractory period by fluctuation in the calculation using the if model @xcite .
the dependence of @xmath86 on @xmath48 is plotted in fig .
[ fige ] for various values of @xmath3 ( @xmath136 , @xmath137 , @xmath119 ) .
it is shown that the synchrony @xmath86 is more increased in smaller system .
the result for @xmath117 is nearly the same as that for @xmath138 . in order to study the dynamical properties of the neuronal ensemble given by eqs .
( [ eq : a1])-([eq : a3 ] ) , we have performed dss by using the heun method @xcite with a time step of 0.0001 : ds results are averages of 100 trials .
amm calculations have been performed for eqs .
( [ eq : g3])-([eq : g5 ] ) by using the fourth - order runge - kutta method with a time step of 0.01 .
we consider , as a typical example , an ensemble with @xmath113 , @xmath139 , @xmath116 , @xmath119 and @xmath117 .
if we adopt a larger @xmath3 in our model calculations , fluctuations in ds results are decreased and we may obtain smoother results , although it needs much computational times . among @xmath49 , @xmath50 and @xmath86 ,
the strongest @xmath3 dependence is realized in @xmath86 [ eqs .
( [ eq : g6])-([eq : g8 ] ) ] .
figure [ fige ] shows that the synchrony @xmath86 almost saturates at @xmath140 where it has little @xmath3 dependence .
this is a reason why we have adopted @xmath117 for our model calculations .
calculated responses to mean- , fluctuation- and synchrony - driven pulse inputs are shown in figs .
[ figf ] , [ figg ] and [ figh ] , respectively , where solid and dashed curves show results of the amm and ds , respectively .
* a. mean - driven inputs * first we apply a mean - driven pulse input given by @xmath141 with @xmath142 , @xmath143 , @xmath144 and @xmath145 .
responses of @xmath54 , @xmath55 , @xmath75 and @xmath146 calculated by the amm ( solid curves ) and ds ( dashed curves ) are shown in figs .
[ figf](a ) , ( b ) , ( c ) and ( d ) , respectively : input signals of @xmath147 , @xmath148 and @xmath46 are plotted by chain curves in ( a ) , ( b ) and ( c ) , respectively . an increase in an applied mean - driven input at @xmath149 induces an increase in @xmath54 and decreases in @xmath55 and @xmath75 which arise from the @xmath150 term in eqs .
( [ eq : g5 ] ) . by an applied pulse input , @xmath146
is decreased because of the increased @xmath49 .
the results of the amm are in fairly good agreement with those of ds .
* b fluctuation - driven inputs * next we apply a fluctuation - driven input given by @xmath151 with @xmath152 , @xmath153 , @xmath154 and @xmath145 . figures [ figg](a)-(d ) show calculated responses of @xmath54 , @xmath55 , @xmath75 and @xmath155 . when the magnitude of @xmath148 is increased at @xmath149 , @xmath55 and @xmath146 are much increased , while there is no changes in @xmath54 .
@xmath75 is modified only at @xmath156 and @xmath157 , where the input pulse is on and off . * c. synchrony - driven inputs * we apply a synchrony - driven input given by @xmath158 with @xmath159 , @xmath160 , @xmath154 and @xmath144 , whose results are plotted in fig . [ figh](a)-(d ) .
an increase in synchrony - driven input at @xmath149 induces increases in @xmath75 , @xmath55 and @xmath146 , but no changes in @xmath54 .
this is because @xmath54 is decoupled from the rest of variables in eqs .
( [ eq : g3])-([eq : g5 ] ) for the case of @xmath161 and @xmath162 .
it is interesting to estimate multivariate input signals from multiple output signals .
such a procedure has been provided in various methods such as bayesian estimation and ica @xcite .
here we consider ica , which was originally developed for a linear mixing system , and then it has been extended to linear and nonlinear dynamical systems .
ica has revealed many interesting applications in various fields such as biological signals and image processing .
a vector @xmath163 of output signals is a real function @xmath164 of a vector @xmath165 of input sources : @xmath166 the dimension of @xmath165 is assumed to be the same or smaller than that of @xmath163 .
if components of @xmath165 are statistically independent and if only one of the source signals is allowed to have a gaussian distribution , ica may extract a vector @xmath167 with a function @xmath168 given by @xmath169 from which we may estimate the original source as @xmath170 @xcite .
* a. coexistence of @xmath47 and @xmath48 * we will discuss the case when mean- and synchrony - driven inputs are simultaneously applied to the neuronal model .
we consider the mean - driven sinusoidal input and synchrony - driven toothsaw input , given by @xmath171 + 0.1 , \label{eq : j3}\\ s_i(t ) & = & 0.01 \:{\rm mod}(t,50 ) , \label{eq : j4}\end{aligned}\ ] ] with @xmath144 where mod(@xmath172 ) denotes the mod function expressing the residue of @xmath102 divided by @xmath173 .
two panels in fig .
[ figj](a ) show input signals of @xmath147 and @xmath46 , and two panels in fig . [
figj](b ) show output signals of @xmath54 and @xmath75 calculated by the amm .
we note a little distortion in @xmath75 due to a cross talk from @xmath54 .
assuming @xmath174 and @xmath175 , we have made an analysis of our result by using ica .
two panels in fig .
[ figj](c ) show two components of @xmath167 extracted from output signals of @xmath175 shown in fig .
[ figj](b ) with the use of the fast ica program @xcite .
although the program is designed for linear , mixing signals , we have employed it for our qualitative discussion .
we note that results in fig .
[ figj](c ) fairly well reproduce the original , sinusoidal and toothsaw signals shown in fig .
[ figj](a ) .
* b. coexistence of @xmath47 , @xmath43 and @xmath48 * next we consider the case where three kinds of inputs are simultaneously applied .
they are mean - driven sinusoidal signal , fluctuation - driven toothsaw signal and synchrony - driven square pulse signal , given by @xmath171 + 0.1 , \label{eq : j5}\\ \gamma_i(t ) & = & 0.002 \:{\rm mod}(t,50 ) , \label{eq : j6}\\ s_i(t ) & = & 0.5 \:\theta[-\cos(2 \pi t/120 ) ] .
\label{eq : j7}\end{aligned}\ ] ] three panels in fig .
[ figk](a ) show the input signals of @xmath147 , @xmath148 and @xmath46 .
output signals of @xmath54 , @xmath55 and @xmath75 calculated in the amm are shown in three panels of fig .
[ figk](b ) , where @xmath55 and @xmath75 are a little distorted by a cross talk .
we have made an ica analysis of our result , assuming @xmath176 and @xmath177 .
three panels of fig .
[ figk](c ) show signals extracted by ica .
extracted sinusoidal and square signals in fig .
[ figk](c ) are similar to those of input signals shown in fig .
[ figk](a ) , though the fidelity of a toothsaw signal is not satisfactory .
this is partly due to the fact that the fast ica program adopted in our analysis is developed for linear mixing models , but not for dynamical nonlinear models @xcite .
these ica analyses show that the mean , fluctuation and synchrony may independently carry information in our population rate - code model .
various attempts have been proposed to obtain the firing - rate model , starting from spiking neuron models @xcite-@xcite .
it is difficult to analytically calculate the firing rate based on the firing model , except for the if - type model @xcite-@xcite .
calculations with the use of the if model have shown the followings : \(1 ) increased input firing rate deceases output variability @xcite , \(2 ) increased input firing rate decreases synchrony @xcite , \(3 ) increased input fluctuation raises firing rate @xcite , \(4 ) increased input synchrony increases firing rates @xcite , and \(5 ) increased input synchrony increases variability @xcite . the items ( 1 ) , ( 2 ) and ( 5 ) are consistent with our result shown in figs .
[ figf ] and [ figg ] .
in contrast , items ( 3 ) and ( 4 ) seem to inconsistent with our result showing that @xmath54 is independent of @xmath148 and @xmath46 , as given by eqs .
( [ eq : g3 ] ) .
it is noted , however , that the model calculation given in the preceding section has been made for the case of @xmath161 and @xmath162 , in which a motion of @xmath54 is decoupled from those of @xmath55 and @xmath75 .
this is not the case in general .
we note in eq .
( [ eq : d6 ] ) that @xmath54 is not decoupled from @xmath55 and @xmath75 except for the case in which ( @xmath96 or 1 ) and ( @xmath98 , @xmath99 or 1 ) .
for example , in the case of @xmath161 and @xmath178 , equations of motion given by eqs .
( [ eq : d6])-([eq : d8 ] ) become @xmath179 equations ( [ eq : l1])-([eq : l3 ] ) clearly show that @xmath54 is coupled with @xmath55 and @xmath75 .
figure [ figl ] shows time courses of @xmath54 , @xmath55 and @xmath75 calculated with the use of eqs .
( [ eq : l1])-([eq : l3 ] ) when a mean - driven input @xmath147 given by eq .
( [ eq : h1 ] ) with @xmath142 , @xmath180 , @xmath181 and @xmath182 is applied to a neuron ensemble with @xmath183 , @xmath116 , @xmath119 , @xmath114 and @xmath117 .
we note that @xmath55 is modified by an applied @xmath147 at @xmath149 , showing the coupling between @xmath54 and @xmath55 in the case of @xmath161 and @xmath178 , while there is no couplings between them in the case of @xmath161 and @xmath162 as shown in fig .
[ figf ] .
a number of neuronal experiments have not reported systematic a change in firing rates while the synchronization within an area is modulated @xcite .
in particular , the synchrony is modified without a change in firing rate in some experiments @xcite .
it has been pointed out that such phenomenon may be accounted for by a mechanism of a rapid activation of a few selected interneurons @xcite . recently
the absence of a change in firing rate is shown to be explained if the ratio of excitatory to inhibitory synaptic weights of long - range couplings is kept constant in neuron ensembles described by the if model @xcite .
we have discussed the rate code of firings neuron population , studying the responses of the generalized rate - code model @xcite to three kinds of inputs : mean- , fluctuation- and synchrony - driven inputs .
the ica analyses of our results have suggested that mean rate , fluctuation ( or variability ) and synchrony may carry independent information .
it would be interesting to examine this possibility by neuronal experiments using _ in vivo _ or _ in vitro _ neuron ensembles .
one of advantages of our rate - code model given by eqs .
( [ eq : a1])-([eq : a3 ] ) is that we can easily discuss various properties of neuronal ensembles , by changing @xmath10 , @xmath11 , @xmath30 and model parameters .
it is possible to theoretically examine various cases of @xmath10 and @xmath11 in a systematic way .
we hope that our rate - code model shares advantages with phenomenological neuronal models such as the hopfield @xcite and wilson - cowan models @xcite .
the tsallis and fisher information entropies are very important quantities expressing information measures in nonextensive systems @xcite .
it is challenging to calculate information entropies in neuronal ensembles with spatially correlated variability @xcite .
the plasticity of synapses ( depression and facilitation ) is known to play important roles in the activity of neurons .
indeed , a memory in brain is accounted for by the plasticity of synapses in the hopfied model @xcite . a variety of activity in prefrontal cortex in response to sensory stimuli
is expected to be explained by dynamic synapses .
it is promising to take into account dynamic synapses in our approach , which is left as our future study .
this work is partly supported by a grant - in - aid for scientific research from the japanese ministry of education , culture , sports , science and technology . the fokker - planck equation ( fpe ) for the langevin equation given by eq .
( [ eq : a1 ] ) in the stratonovich representation is given by @xcite @xmath184 \:p(\ { r_{k } \},t)\ } \nonumber \\ & & + \frac{\beta^2}{2}\sum_{i}\sum_{j } \frac{\partial^2}{\partial r_{i } \partial r_{j } } \ { [ \delta_{ij}+c_0 ( 1-\delta_{ij } ) ] \:p(\ { r_{k } \},t ) \ } , \nonumber \\ & & + \frac{\alpha^2}{2}\sum_{i } \sum_j [ \delta_{ij}+c_1(1-\delta_{ij } ) ] \frac{\partial}{\partial r_{i } } \ { g(r_i ) \frac{\partial}{\partial r_j } [ g(r_{i } ) \:p(\ { r_{k } \},t ) ] \}.
\label{eq : x1}\end{aligned}\ ] ] equations of motion for moments , @xmath185 and @xmath186 , are derived with the use of fpe @xcite : @xmath187
\rangle + \langle r_j\ : [ f(r_i)+ h(u_i ) ] \rangle \nonumber \\ & + & \frac{\alpha^2}{2 } [ \langle r_i g'(r_j ) g(r_j ) \rangle + \langle r_j g'(r_i ) g(r_i)\rangle ] \nonumber \\ & + & \delta_{ij } [ \gamma_i+\alpha^2\:\langle g(r_i)^2 \rangle + \beta^2 ] \nonumber \\ & + & ( 1-\delta_{ij } ) [ \gamma_i s_i + c_1 \alpha^2 \langle g(r_i)g(r_j ) \rangle+c_0 \beta^2 ] .
\label{eq : x3}\end{aligned}\ ] ] we may obtain eqs .
( [ eq : b6])-([eq : b8 ] ) , by using the expansion given by @xmath188 and the relations given by @xmath189 for example , eq . ( [ eq : b6 ] ) for @xmath190 is obtained as follows .
@xmath191 equations ( [ eq : b7 ] ) and ( [ eq : b8 ] ) are obtainable in a similar way . by using eq .
( [ eq : c2 ] ) , we obtain equations of motion for @xmath54 , @xmath55 and @xmath75 given by eqs .
( [ eq : d1])-([eq : d3 ] ) . in making a linear stability analysis
, it is better to adopt the basis of @xmath192 than that of @xmath88 . in the former basis ,
equations of motion for @xmath161 and @xmath162 given by eqs .
( [ eq : b6])-([eq : b8 ] ) become @xmath193 the stability of the stationary state may be examined by the jacobian matrix of eqs .
( b1)-(b3 ) . with the use of @xmath194 in the matrix
, we obtain its eigenvalues given by @xmath195 from which the stability condition given by eqs .
( [ eq : g11 ] ) and ( [ eq : g12 ] ) is obtained .
a. riehle , s. grn , m. diesmann , and a. aertsen , science * 278 * ( 1997 ) 1950 .
s. c. de oliveira , a. thiele , and k. p. hoffman , j. neurosci .
* 17 * ( 1997 ) 9248 .
p. fries , j. h. reynolds , a. e. rorie , and r. desimine , science * 291 * ( 2001 ) 1560 .
a. d. reyes , nat .
* 6 * ( 2003 ) 593 .
f. grammont and a. riehle , biol . cybern .
* 88 * ( 2003 ) 360
. p. h. e. tiesinga and t. j. sejnowski , neural comput .
* 16 * ( 2004 ) 251
. f. s. chance , l. f. abbott , and a. d. reyes , neuron * 35 * ( 2002 ) 773 .
g. silberberg , m. bethge , h. markram , k. pawelzik , and m. tsodyks , j. neurophysiol . * 91 * ( 2004 ) 704 .
m. arsiero , h - r .
lscher , b. n. lndstrom , and m. giugliano , j. neurosci .
* 27 * ( 2007 ) 3274 .
p. h. e. tiesinga , j. v. jos , and t. j. sejnowski , phys .
e * 62 * ( 2000 ) 8413 . for a review on calculations using the if model , see : a. n. burkitt , boilogical cybernetics * 95 * ( 2006 ) 1 ; * 95 * ( 2006 ) 97 .
b. lindner and l. schimansky - geier , phys .
* 86 * ( 2001 ) 2934 .
r. moreno , j. de la rocha , a. renart , and n. parga , phys .
* 89 * ( 2002 ) 288101 .
a. longtin , b. doiron , and a. r. bulsara , biosystems * 67 * ( 2002 ) 147 .
b. lindner , a. longtin , and a. bulsara , neural comput .
* 15 * ( 2003 ) 1761 .
a. n. burkitt , h. meffin , and d. b. grayden , biol . cybern . * 89 * ( 2003 ) 119 .
a. kuhn , a. aertsen , and s. rotter , j. neurosci .
* 24 * ( 2003 ) 2345 .
m. schindler , p. talkner , and p. hnggi , phys , rev , lett .
* 93 * ( 2004 ) 048102 .
b. doiron , j. rinzel , and a. reyes , phys .
e * 74 * ( 2006 ) 030903(r ) .
n. masuda , neural comput .
* 18 * ( 2006 ) 45 .
a. renart , r. moreno - bose , x - j .
wang , and n. parga , neurla comput . * 19 * ( 2007 ) 1 .
j. heinzle , p. knig , r. f. salazar , cogn . neurodyn .
* 1 * ( 2007 ) 225 .
h. hasegawa , phys .
rev e * 75 * ( 2007 ) 051904 . in the heun method for the ordinary differential equation of @xmath196 , a value of @xmath197 at @xmath198
is evaluated by @xmath199 $ ] with @xmath200 and @xmath201 , while it is given by @xmath202 in the euler method , @xmath203 being the time step .
the heum method for the stochastic ordinary differential equation meets the stratonovich calculus employed in the fpe @xcite .
w. rmelin , siam j. numer .
19 * ( 1982 ) 604 ; a. greiner , w. strittmatter , and j. honerkamp , j. stat .
* 51 * ( 1988 ) 95 ; r. f. fox , i. r. gatland , r. roy , and g. vemuri , phys .
a * 38 * ( 1988 ) 5938 .
the fast ica program is available at http://www.cis.hut.fi / projects / ica/. h. hasegawa , phys .
e * 77 * ( 2008 ) 031133 .
h. hasegawa , arxiv:08xx.xxxx ( in preparation ) .
h. haken : _ advanced synergetics _
( springer - verlag , berlin , 1983 ) .
m. ibanes , j. garcia - ojalvo , r. toral , and j. m. sancho , phys .
e * 60 * ( 1999 ) 3597 . | a population of firing neurons is expected to carry information not only by mean firing rate but also by fluctuation and synchrony among neurons . in order to examine this possibility ,
we have studied responses of neuronal ensembles to three kinds of inputs : mean- , fluctuation- and synchrony - driven inputs . the generalized rate - code model including additive and multiplicative noise ( h. hasegawa , phys .
rev .
e * 75 * ( 2007 ) 051904 ) has been studied by direct simulations ( dss ) and the augmented moment method ( amm ) in which equations of motion for mean firing rate , fluctuation and synchrony are derived .
results calculated by the amm are in good agreement with those by dss .
the independent component analysis ( ica ) of our results has shown that mean firing rate , fluctuation ( or variability ) and synchrony may carry independent information in the population rate - code model .
the input - output relation of mean firing rates is shown to have higher sensitivity for larger multiplicative noise , as recently observed in prefrontal cortex .
a comparison is made between results obtained by the integrate - and - fire ( if ) model and our rate - code model .
* population rate codes carried by mean , fluctuation + and synchrony of neuronal firings * hideo hasegawa _ department of physics , tokyo gakugei university , + koganei , tokyo 184 - 8501 , japan _ ( ) _ pacs no . _ : 87.10.+e , 84.35.+i _ keywords _ : neuron population , rate - code model , firing rate , synchrony , variability |
in a series of experiments @xcite , a nonuniform bose - einstein condensate ( bec ) of @xmath0li atoms was formed and proved metastable for atom numbers @xmath1 with @xmath2 , in agreement with theoretical predictions @xcite . due to attraction between atoms ,
such condensate is bound to collapse when @xmath3 , but it may collapse also for @xmath1 via quantum or thermal tunneling . within the mean - field appraoch , bec is described by one wave function .
our aim in this work was to find the exact mean - field description of the quantum tunneling in this simplest conceivable many - body system .
our additional motivation is the recently demonstarted ability to control the interaction of @xmath4rb atoms in bec @xcite , which opens a new perspective to systematic experimental checks on quantum tunneling . up to now , studies of the quantum tunneling of bec relied on assuming gaussian wave functions @xcite , at least in assigning the mass parameter @xcite .
strictly speaking , once the mean field equation is specified , there is no place for such an assumption : this equation , taking a form of the nonlinear field equation , by itself determines the dynamics . to find solutions
which correspond to quantum decay we use the method of instantons , i.e. fields evolving in imaginary time @xcite , carried over to mean - field theories of many - body systems @xcite .
it gives the decay rate @xmath5 , with the exponent @xmath6 being the action for the optimal mean - field instanton , called bounce .
we find the exact instantons for spherical bec by first transforming the original imaginary - time mean - field equation to an equation for the condensate density , and then solving it numerically . having done that wa can check the gaussian ansatz .
the real - time version of the obtained equation encompasses finite amplitude collective radial oscillations of bec .
applying quantization rule we find energies of radial eigenmodes . by using imaginary - time dynamics
we also find periodic instantons determining decay exponents of these breathing modes . in this way
, the collective dynamics of an attractive spherical condensate close to instability is obtained from the mean field equation .
we assume that the dynamics of bec is governed by the time - dependent gross - pitaevskii equation ( gpe ) @xcite @xmath7 where @xmath8 is the static trap potential and @xmath9 , with @xmath10 the s - wave scattering length and @xmath11 the atomic mass .
the wave function is normalized as @xmath12 , with @xmath13 the total number of atoms in the condensate . in the present work we consider a harmonic , spherically symmetric trap @xmath14 .
this suggests choosing @xmath15 as the unit of length , @xmath16 as the unit of time and @xmath17 as the unit of energy .
we also change the normalization of the wave function to unity , @xmath18 . to simplify equations we work with the function @xmath19 , for which the gpe reads @xmath20 where @xmath21 , and the generalised density @xmath22 .
for a stationary state , @xmath23 , @xmath24 being the single - particle energy or chemical potential .
for each @xmath25 there are two stationary states , one metastable and another unstable , at the top of the energy barrier @xcite ( cf fig .
quantum tunneling of bec is described by a specific solution to the equation @xmath26 obtained from ( [ gp1 ] ) by a transition to imaginary time , @xmath27 @xcite .
now , the density @xmath28 as @xmath29 is replaced by @xmath30 upon @xmath27 .
this makes eq.([eqsp ] ) nonlocal in time .
bounce has to satisfy the boundary conditions of 1 ) periodicity , @xmath31 , with @xmath32 the amplitude of the metastable state , and 2 ) barrier penetration , i.e. @xmath33 has to be some state of bec at the other side of the potential barrier .
( [ eqsp ] ) preserves both the normalization @xmath34 and the energy @xmath35 with @xmath36 the energy of the metastable state . for a bounce starting from the metastable state the period @xmath37 extends to infinity @xcite .
the decay exponent reads @xcite @xmath38 since eq.([eqsp ] ) is real and the boundary value @xmath39 may be taken real , we assume real @xmath40 in the following .
now the point is to transform the nonlocal in time instanton equation ( [ eqsp ] ) into an evolution equation for the condensate density .
a transformation of the real - time gpe ( [ gp1 ] ) to a fluid - dynamic form provides an analogy , but is conceptually simpler .
the bounce equation ( [ eqsp ] ) , with the boundary conditions specified above , splits into two time - local equations for the time - even density @xmath41 and the time - odd current @xmath42 @xmath43 @xmath44 = 0 , \ ] ] where the kinetic energy density @xmath45/\rho$ ] .
when @xmath41 is non - negative ( which is very plausible in the present case , but not granted in general as @xmath46 not @xmath47 ) , one can define a regular , time - odd function @xmath48 which allows decomposition @xmath49 . from this , the fluid - dynamic representation of ( [ eqsp ] ) follows , with the velocity field @xmath50 .
however , even for arbitrary @xmath41 , one can eliminate @xmath51 from eqs.([roj1],[roj2 ] ) , which is more convenient .
introducing @xmath52 , so that @xmath53 and @xmath54 , we automatically fulfil ( [ roj1 ] ) , and ( [ roj2 ] ) transforms to the equation for the primitive of the bounce density , basic for the imaginary - time dynamics of spherical bec - s : @xmath55 = 0 .\ ] ] notice , that the finite amplitude oscillations of bec around the metastable minimum are described by the real - time version of eq.([bou ] ) , in which @xmath56 and @xmath57 are replaced by @xmath58 and @xmath59 , respectively .
an alternative , global approach to quantum tunneling is to look for a minimum of action ( [ s ] ) under the condition of constant energy ( [ mhf ] ) and norm . indeed , for a regular @xmath60 , i.e. positive @xmath41 , @xmath61 .
using eq.([mhf ] ) and introducing an observable @xmath62 uniquely labelling states along the barrier , explicitly @xmath63 , we obtain the following functional @xmath64=2 n\int_{q(0)}^{q(\tau_p/2 ) } dq \sqrt{2 b(q ) ( v(q)-{\cal e})}\ ] ] which is minimized by the primitive of the bounce density ( note , that @xmath65 for bec ) .
the potential energy @xmath66 $ ] , @xmath67=4\pi\int_0^{\infty } dr[\frac{(\partial_r\rho)^2}{8\rho}+\frac{1}{2}\rho r^2+\frac{k\rho^2}{2 r^2}]\ ] ] and the effective mass parameter @xmath68 $ ] @xmath69 are both the functionals of @xmath70 .
note that eq.([func ] ) is invariant with respect to a change of the controlling variable , as for any other such variable @xmath71 , @xmath72 .
the energy conservation ( [ mhf ] ) implies that for bounce @xmath73 , with @xmath74 , and therefore @xmath75 $ ] .
we have solved eq.([bou ] ) using the variable @xmath62 rather than @xmath76 .
an initial sequence of densities @xmath77 , with 30 - 50 @xmath78 points covering the barrier region , was constructed by minimizing @xmath79 $ ] ( [ v ] ) under the constraint @xmath80 .
these constrained stationary densities were then improved upon iteratively .
the details of @xmath81 , the counterpart of @xmath82 of eq .
( [ eqgp ] ) , are obtained more precisely when the @xmath83 behaviour near @xmath84 and the harmonic oscillator asymptotics at infinity are factored out in eq.([bou ] ) . in numerical work ,
we use a function @xmath85 such that @xmath86 , and properly express the term @xmath87 - see _ appendix_.
we also performed the minimization of the functional ( [ func ] ) treating @xmath88 as independent variables .
it turns out that @xmath89 thus obtained do not fulfil eq.([bou ] ) accurately , but the accuracy in @xmath6 is better than 0.1% .
the numerical results have been obtained with physical data on @xmath0li adopted after the most accurate treatment up to date @xcite .
these values give @xmath90 , and we obtain the critical value @xmath91 between 7.2249 and 7.2255 ( @xmath92 between 1258.7 and 1258.8 ) .
the potential energy @xmath93 from the bounce solutions ( fig.1 ) , nearly identical with @xmath94 $ ] for constrained stationary @xmath77 , is very flat between @xmath95 and @xmath96 for @xmath97 . for smaller @xmath13
, it becomes quite peaked around the summit , and its fall on the side of small @xmath62 becomes very steep .
all the obtained bounce solutions result from the small adjustment of the initial densities . for larger @xmath98 ( and increasing @xmath99 and @xmath100 terms , cf eq.([edyn ] ) ) this adjustment becomes gradually more difficult .
we could not obtain the exact solution of eq.([bou ] ) ( or even a constrained stationary @xmath77 for small @xmath78 ) for @xmath101 .
the mass parameters @xmath102 from various instanton solutions ( fig.2 ) are nearly identical which shows that there exists a universal inertia for the radial collective motion of bec close to instability .
this may be understood as a consequence of a nearly static character of solutions @xmath103 in the limited range of @xmath104 values of interest : @xmath70 depends weakly on @xmath13 close to critical @xmath92 . for a gaussian density with a variable width @xmath105 , @xmath106 , the current @xmath107 ( eq.([roj1 ] ) ) , and the mass parameter @xmath108 ( cf @xcite ) .
hence , as for the gaussian density @xmath109 , one has @xmath110 , as used in ref.@xcite . as seen in fig.2 , this is a fair assumption near the metastable minima , much worse though for smaller @xmath62 around the barrier s summit . the error in @xmath6 due to the gaussian value of @xmath111 amounts to 3 - 4% in the cases studied .
the bounce `` amplitude '' , @xmath112 ( fig.3 ) , differs from the constrained stationary values @xmath113 mostly near @xmath84 and for small @xmath62 , by up to 1% .
decay exponents calculated with the initial densities @xmath77 are only up to 0.3% larger than the exact ones .
next we turn to collective radial oscillations of bec and their decay rates .
quantum tunneling from the excited state with energy @xmath114 may be treated by solving eq.([bou ] ) as before , except that now the period @xmath37 must be finite and @xmath115 replaced by @xmath116 .
one has to fix the boundary values @xmath117 and the energies @xmath114 .
the energies @xmath118 of the excited quantum states follow from the quantization condition @xmath119 @xcite , where @xmath120 , cf eq.([func ] ) , with the period @xmath121 depending on @xmath118 .
essentially , the imaginary - time boundary value @xmath117 should match the real - time oscillatory solution @xmath122 at the turning point @xmath62 at which @xmath123 .
we could not solve the real - time version of eq.([bou ] ) as accurately as for instantons .
however , since the action @xmath6 is insensitive to the details of solution , the lowest quantized @xmath118 are quite accurate .
their comparison to frequencies of the small amplitude oscillations , obtained from the gpe linearized around @xmath124 ( so called bogolyubov spectrum , see @xcite ) , is interesting .
the first radial excitation , @xmath125 , is nearly equal , but little smaller than , the lowest bogolyubov mode corresponding to the small amplitude limit , except that it may not exist for a too shallow @xmath93 , like for @xmath126 .
the higher modes , @xmath118 , @xmath127 , come out roughly at the multiples of @xmath125 ( table 1 ) , and thus represent nearly harmonic spectrum of collective radial oscillations .
its @xmath128 states are lower than the second bogolyubov mode , lying slightly above @xmath129 . using the quantized oscillation energies we have found periodic instantons and calculated decay exponents for some low collective radial excitations .
such instanton with a period @xmath37 is quite similar to the so - called `` thermon '' which describes thermal decay from the metastable ground state at the temperature @xmath130 @xcite .
this similarity is easily understood : thermal decay goes via thermal excitation of quantum states above the metastable ground state and successive tunneling from them through the smaller barrier . at higher temperatures , details of the excitation spectrum and tunneling
become irrelevant and the thermal tunneling rate @xmath131 , with @xmath132 the barrier energy , replaces a thermal mixture of rates @xmath133 from a few lowest excited states at small temperature .
the critical temperature @xmath134 at which this happens is usually determined as the one at which @xmath135 .
numerical results are collected in table 1 .
those in columns 2 - 4 depend only on @xmath136 and relate to all bec - s with attractive interaction .
decay rates @xmath137 from the metastable @xmath0li bec were calculated for @xmath138 s@xmath139 @xcite , assuming prefactor @xmath140 @xcite , with @xmath141 from the small amplitude limit .
the exact prefactor is difficult to calculate , but it must be of the same order as in table 1 . obtained values of @xmath137 are very close to those of ref.@xcite , where the effect of too small @xmath111-s is fortuitously reduced by half by the interpolation overestimating @xmath93 .
the decrease of the decay exponents @xmath142 with the mode number @xmath143 and their comparison to exponents @xmath6 for the ground state are seen in table 1 , col .
5 . for example
, quantum decay of the @xmath144 radial mode in bec with @xmath145 is nearly as quick as for the ground state of the @xmath126 bec .
the crossover temperature @xmath134 at which the thermal decay begins to dominate over the quantum tunneling is approximated , neglecting prefactors , as @xmath146 , with @xmath132 the barrier energy , @xmath36 for the metastable state , and @xmath147 for the excited state . for the latter ,
the meaning of critical temperature is extended in analogy with that for the ground state : suppose the condensate in the oscillation state @xmath143 and ask at which temperature @xmath148 the thermal rate exponent equals the instanton action @xmath142 .
these ratios ( table 1 , column 3 ) show that for presently measurable @xmath137 and @xmath138 s@xmath139 , corresponding to @xmath149 nk , @xmath150nk , both for ground state and collective radial oscillations .
we notice , that for all finite-@xmath37 instantons describing decay out of the @xmath151th collective radial excitation , @xmath152 so , at their own thermon `` temperature '' , they dominate over the thermal decay .
the quantity @xmath153 ( table 1 , column 4 ) shows @xmath154 behaviour close to @xmath92 , with @xmath155 @xcite . for larger @xmath98
the effective exponent slightly decreases to @xmath156 .
consider now two different attractive bec - s with critical particle numbers @xmath92 and @xmath157 .
the same decay exponent @xmath6 is obtained for such ( not too large ) @xmath98 and @xmath158 which satisfy the relation @xmath159 , with @xmath160 .
thus , e.g. we obtain @xmath161 for @xmath162 in bec with @xmath163 , and for @xmath164 in bec with @xmath165 , cf table 1 . in summary ,
the equations for the condensate density , describing both the real- and imaginary - time dynamics of spherical bec , were formulated and the exact instanton solutions were found numerically , also for collective radial excitations .
the determined mass parameter ( fig .
2 ) deviates from the gaussian ansatz , but the calculated decay exponents for the metastable states agree well with ref.@xcite .
it follows from fig.2 that the exact mass parametr may be more relevant to the behind - the - barrier collapse phase , i.e for smaller @xmath62 .
the quantized energies of collective finite amplitude radial vibrations form nearly harmonic ( slightly compressed ) spectrum with @xmath166 , where @xmath141 is slightly lower than the lowest bogolyubov mode .
the @xmath128 collective oscillation states lie lower than the second bogolyubov mode .
any excitation ( thermal or otherwise , e.g. by modulation of the trapping oscillator frequancy ) of these collective vibrations must lead to a faster decay of the condensate , as the table 1 shows .
if quantum tunneling is not to be overshadowed by thermal decay , the experiments should proceed at low @xmath167 and/or large @xmath168 .
since theoretical results are more certain away from @xmath92 , where the exponent @xmath6 dominates decay , one should probe a range of moderate @xmath6 , giving observable , but not too large @xmath137 ( perhaps , by discarding prompt collapses ) .
the corresponding range of @xmath98 values depends on @xmath92 as @xmath169 .
we have @xmath170 with @xmath171 .
the stationary gpe leads to the equation for @xmath172 @xmath173 with @xmath174 . for large @xmath175 , @xmath176 @xcite .
the normalization of @xmath177 implies the next two terms @xmath178 a solution regular at @xmath179 must fulfil @xmath180 these boundary conditions suggest a method of solution : for a given @xmath136 we assume some @xmath24 and @xmath181 and , starting from the asymptotic values ( [ eas ] ) at large @xmath175 , integrate eq.([estat ] ) to @xmath179 .
we check eq.([eor ] ) and the normalization and correct @xmath24 and @xmath181 until we fulfil both . by using a new variable @xmath182 and factoring out @xmath183 , we transform eq.([bou ] ) to a form @xmath184)= \frac{\partial r}{\partial s } , \ ] ] suitable for both instantons and oscillations if @xmath185 is understood
. @xmath186 $ ] stands for the l.h.s . of eq.([estat
let us call the l.h.s . of eq.([edyn ] ) @xmath187 .
if @xmath188 ( no time dependence ) we recover stationary solutions of eq.([estat ] ) . for instantons ,
we solve eq.([edyn ] ) iteratively . having a set of @xmath189
we calculate @xmath190 . for each @xmath191
we solve ( [ edyn ] ) as the ordinary differential equation in @xmath175 to obtain new @xmath189 .
the method , as for the stationary case , is to adjust the asympotic form ( [ eas ] ) to the proper regularity condition at @xmath179 .
the new and old @xmath191-s are combined to provide initial @xmath191-s for the next iteration . with a careful modification of @xmath191-s
this iteration leads to the self - consistency , i.e. @xmath192 . the initial densities @xmath193 are obtained using the constrained imaginary - time step hartree procedure .
we use the runge - kutta - merson procedure for integration of eqs.([edyn ] ) .
energies are calculated using a mesh of 128 equidistant points , @xmath194 , and the cubic spline interpolation for derivatives .
we have checked that doubling the mesh density does not change results in any appreciable manner .
k. g. singh and d. s. rokhsar , phys .
77 ( 1996 ) 1667 , m. edwards , p. a. ruprecht , k. burnett , r. j. dodd and c. w. clark , phys .
77 ( 1996 ) 1671 , p. a. ruprecht , m. edwards , k. burnett , and c. w. clark , phys .
a 54 ( 1996 ) 4178 table 1 - energies , crossover temperatures , decay exponents and rates for metastable and radially excited states of the @xmath0li bec .
results from bounce solutions ( @xmath195 from the functional minimization ) . | imaginary- and real - time versions of an equation for the condensate density are presented which describe dynamics and decay of any spherical bose - einstein condensate ( bec ) within the mean field approach .
we obtain quantized energies of collective , finite amplitude radial oscillations and exact numerical instanton solutions which describe quantum tunneling from both the metastable and radially excited states of the bec of @xmath0li atoms .
the mass parameter for the radial motion is found different from the gaussian value assumed hitherto , but the effect of this difference on decay exponents is small .
the collective breathing states form slightly compressed harmonic spectrum , n=4 state lying lower than the second bogoliubov ( small amplitude ) mode .
the decay of these states , if excited , may simulate a shorter than true lifetime of the metastable state . by scaling arguments ,
results extend to other attractive becs .
pacs number(s ) : 03.75.fi , 05.30.jp , 05.45.-a , 47.20.ky |
an ultimate goal of the study of quasar ( qso ) absorption lines is to develop a comprehensive understanding of the kinematic , chemical , and ionization conditions of gaseous structures in early epoch galaxies and to chart their cosmic evolution . for a comprehensive physical picture of any given absorption system ,
both high resolution spectra of a wide range of chemical and ionization species and the empirically measured properties of the associated galaxies are required . for @xmath10 , shortly following the epoch of peak star formation , the association between mg0.1emii @xmath11 absorption and galaxies is well established ( @xcite ; @xcite ) , and their kinematics , though complex and varied , are consistent with being coupled to the galaxies themselves ( @xcite ; @xcite ) .
it is unfortunate , however , that for @xmath12 , the spectroscopic data of are not of uniform , high quality due to the need for large amounts of space based telescope time to observe ultraviolet wavelengths .
presently , any comprehensive analyses of low redshift systems for which the low ionization species ( i.e. , , ) have been observed at high resolution with hires / keck ( see churchill , vogt , & charlton 1999b ) must incorporate low resolution fos/_hst _ spectra of the intermediate and high ionization species , especially the strong c0.1emiv @xmath13 , n0.1emv @xmath14 , and o0.1emvi @xmath15 doublets , and of several other important low ionization species .
presently , it is not clear if these low resolution data can be used to place meaningful constraints on the chemical and ionization conditions of the clouds in selected absorbers . in this paper
, we investigated this issue in a pilot study , since an affirmation would imply that a larger sample could be studied using existing data from the _ hst _ archive .
we would then be able to address the broader implications for galaxy formation scenarios based upon the inferred metallicities , abundance patterns , and inferred relative spatial distribution of the low and high ionization absorbing gas clouds . under the assumptions of photoionization and/or collisional ionization equilibrium
, we developed a technique in which it was assumed that the number of clouds and their kinematics are obtained by voigt profile decomposition of the high resolution spectra .
we then used the lower resolution profiles from the fos data to place constraints on the range of chemical and ionization conditions in these clouds .
we also explored the idea that the could arise in relatively low ionization clouds embedded in a higher ionization and more extended medium ( see @xcite ; @xcite ) .
more specifically , we set out to answer three questions : ( 1 ) assuming the clouds measured with hires are photoionized , can we construct model clouds that are consistent with the many low and intermediate ionization species captured in the fos data ? ( 2 ) if so , are we required to infer an additional ( presumedly low density and diffuse ) component to account for higher ionization absorption from , , and ? ( 3 ) if so , can this diffuse component be made consistent with photoionized only , collisionally ionized only , or photo plus
collisionally ionized gas ? for this paper , we chose the three systems at @xmath16 , @xmath3 , and @xmath4 along the line of sight toward pg @xmath5 ( @xmath17 ) because they are exceptionally rich in low , intermediate , and high ionization ultraviolet transitions ( @xcite ; @xcite ; @xcite ) .
the hires / keck profiles are illustrated in figure [ fig : order28 ] .
two of the systems are kinematically `` complex '' and are separated by @xmath18 .
the third system is more isolated , being @xmath19 from the other two .
this system is classified as a `` weak '' absorber [ defined by @xmath20 ( @xcite ) ] .
the highest ionization transitions , , and , are seen to have a total kinematic spread of @xmath21 coincident with the three system seen in absorption ( @xcite ) . in the qso field ,
kirhakos et al .
( 1994 ) found three bright galaxies with angular separations from the quasar of 5.6 , 8.6 , and 9.0 and @xmath22 magnitudes 21.1 , 21.5 , and 22.3 , respectively .
the 8.6 galaxy has detected @xmath23 with flux @xmath24 ergs s@xmath25 at @xmath26 ( @xcite ) . at this redshift , the qso
galaxy impact parameters are @xmath27 , @xmath28 , and @xmath29 kpc ( @xmath30 )
. there are @xmath31 galaxies with @xmath32 within 100 of the qso ( @xcite ) .
this is an overdensity by a factor of @xmath33 compared to field galaxies ( @xcite ) .
thus , it is of interest to entertain the possibility of a group environment for these absorbers . in
[ sec : data ] we describe the data and its analysis . in
[ sec : models ] we outline our modeling technique and simplifying assumptions .
a synopsis of the model results are given in
[ sec : results ] .
details on how the data were used to constrain the models and how various ionizing spectral energy distributions modify these models are given in appendices [ app : constraints ] and [ app : models ] . in
[ sec : discussion ] , we compare and contrast the system properties , and in [ sec : onthe ] , we discuss what might be inferred about the relative spatial distribution of the low and high ionization gas .
we summarize in [ sec : conclusion ] .
the optical data were obtained with the hires spectrometer ( @xcite ) on the keck i telescope on 23 january 1995 ut under clear and stable conditions with a seeing of @xmath34 .
the spectral resolution is @xmath35 ( @xmath36 ) , with a sampling of 3 pixels per resolution element .
the signal to noise ratio is @xmath37 per resolution element .
the @xmath38 , @xmath39 , @xmath40 , @xmath41 , and @xmath42 transitions were captured at similar signal to noise ratio .
the hires spectrum was reduced with the iraf _ apextract _ package for echelle data .
the detailed steps for the reduction are outlined in churchill ( 1995 ) .
the spectrum was extracted using the optimized routines of horne ( 1986 ) and marsh ( 1989 ) .
the wavelengths were calibrated to vacuum using the iraf task _ ecidentify _ , which models the full 2d echelle format .
the absolute wavelength scale was then corrected to heliocentric velocity .
the continuum normalization was performed using the iraf _ sfit _ task .
objective and unbiased identification of absorption features ( without regard to their association with the studied systems ) was performed as described in churchill et al .
( 1999b ) , using the methodology of schneider et al .
( 1993 ) .
three systems , which we hereafter call a , b , and c , are observed at redshifts @xmath43 , @xmath44 , and @xmath45 . in figure
[ fig : order28 ] , we present the doublet profiles .
the doublets are marked by the labeled bar above the normalized continuum .
the solid curve through the data is a model spectrum generated using voigt profile ( vp ) decomposition .
the free parameters are the number of vp components ( `` clouds '' ) and , for each , its redshift , column density , and doppler @xmath46 parameter .
we have used the program minfit ( @xcite ) , which performs a @xmath47 minimization while minimizing the number of clouds using a specified confidence level and the standard f test .
the adopted vp decomposition had six clouds in system a , five clouds in system b , and a single cloud in system c. the hires data and the vp decompositions are shown in figure [ fig : vpfits ] with the profiles aligned in rest frame velocity . in table
[ tab : vpfits ] , we present the cloud properties , including individual redshifts , velocities with respect to @xmath48 , column densities , @xmath46 parameters , and the ratios @xmath49 .
only system b was found to have measurable .
the @xmath50 upper limits on were obtained for each cloud from the equivalent width limits of the @xmath51 transition . for in system a , we measured a mean @xmath50 column density upper limit of @xmath52 for the six clouds using the technique of stacking ( @xcite ) .
the ultraviolet data were obtained with the faint object spectrograph ( fos ) on the _ hubble space telescope _ as part of the qso absorption line key project .
the data acquisition , their reduction , and the objective absorption line lists are presented in jannuzi et al .
their fully reduced g190h and g270h spectra have kindly been made available for this study .
the spectra have a resolution of @xmath53 ( @xmath54 ) and cover the approximate wavelength intervals @xmath55 to @xmath56 and @xmath57 to @xmath58 for the g190h and g270h settings , respectively .
for the most part , we have adopted the key project continuum fits and line identifications .
some refinement was needed near the si0.1emiv @xmath59 doublet ( @xmath60 on g270h ) , which lies on the red wing of the broad emission line , and near the si0.1emii @xmath61 doublet ( @xmath62 on g190h ) , which lies at the spectrum edge .
we have re fit the continuum across these regions .
a lyman limit break is present at @xmath63 ( however , see stengler larrea 1995 ; jannuzi et al .
we extrapolated the continuum fit of jannuzi et al . below the break starting at the break shoulder ( @xmath64 ) .
this technique preserved the measured optical depth , or the break ratio , @xmath65 ( @xcite ) , while yielding a reasonable approximation to the shape of the recovery ( see figure 3@xmath66 ) . due to the high density of lines ,
the jannuzi et al .
continuum may have been systematically low by 510% in the region 1790 to 1820 and our extrapolation may have propagated this systematic offset .
nonetheless , this error has a negligible effect on the measured break ratio of @xmath67 , which we obtained from the unnormalized as well as the normalized spectrum .
this ratio implies @xmath68 and a total neutral hydrogen column density of @xmath69 . out of the @xmath70 qsos analyzed by the qso absorption line key project ,
pg @xmath5 is one of eight for which the line identifications ( ids ) were subject to greater uncertainty ( the forest is ubiquitous blueward of 2623 and at least four metal systems are present ) . using the hires data for cross checks with the jannuzi et al . line ids in the fos spectrum , we have made a table of transitions , including those detected and those used for constraining the models . in table [ tab : fosids ] ( end of paper ) , we have listed the observed wavelengths , line ids , ionization potentials , and notes on any blending .
for example , we identified the lyman series down to , beyond which the three systems blend together .
the fos spectrum provides a large number of chemical and ionization species , including , , , , , @xmath71 and @xmath72 , @xmath73 , c0.1emiv @xmath13 , @xmath74 , @xmath75 , n0.1emv @xmath14 , o0.1emvi @xmath15 , si0.1emii @xmath61 and @xmath76 , @xmath77 , si0.1emiv @xmath59 , and s0.1emvi @xmath78 .
these species represent a wide range of ionization potentials from 13.6 ev for to 138.1 ev for .
assuming the clouds are photoionized , our goal was to determine if we can obtain useful constraints on their range of chemical and ionization conditions using the fos/_hst _ spectra . within this context , we also explored the possibility of a high ionization ( , , and ) , presumedly diffuse , component . in principle , this high ionization phase could be photoionized , collisionally ionized , or photo plus collisionally ionized gas ( a spatially segregated two phase high ionization component ) . for both the clouds and the high ionization phase , the extragalactic ultraviolet ionizing background spectrum of haardt & madau ( 1996 ) for @xmath79 was assumed . using the @xmath23 detection and constraints measured by thimm ( 1995 )
, we explored the range of allowed contributions ( modifications to the haardt & madau background ) from various galactic spectral energy distributions .
we find that galactic contributions , within the allowed ranges explored , do not modify _ general _ conclusions based upon the assumption of a pure haardt & madau background ( see appendix [ app : models ] ) .
we used the photoionization code cloudy ( version 90.4 ; @xcite ) .
the free parameters are the neutral hydrogen column density , @xmath80 , the metallicity } = \log ( { \rm x / y } ) - \log ( { \rm x / y})_{\odot}$ ] and for metallicity use @xmath81 . ] , @xmath82 , the abundance pattern , and the ionization `` parameter '' , @xmath83 , which is defined as the ratio of the number of hydrogen ionizing photons to the hydrogen number density , @xmath84 ( including ionized , neutral , and molecular forms ) . for the haardt & madau spectrum and normalization , a simple relation between ionization parameter and hydrogen number density , @xmath85 , holds . for a given abundance pattern ,
the ratio @xmath86 uniquely determines the ionization parameter , @xmath83 . once @xmath83 is determined , the measured @xmath87 fixes @xmath88 , for a cloud in photoionization equilibrium . for a given @xmath82 ,
the constant @xmath89 is constrained by the lyman series transitions in the fos data . throughout , we assume @xmath90 . to characterize the abundance pattern , we used the ratio of @xmath91group species to fe group species , [ @xmath91/fe ] .
abundance ratios measured in galactic stars show a clear range of @xmath92 \leq + 0.5 $ ] ( @xcite ) , which we adopt as a reasonable range for the studied systems .
since all clouds have measured @xmath87 , an @xmath91group element , it follows that @xmath93 + z \simeq c_{2}$ ] .
thus , for clouds with measured @xmath86 , the only arbitrarily chosen free parameter is the abundance pattern .
one selects a @xmath82 and [ @xmath91/fe ] ( fixes @xmath94 ) , determines @xmath80 by constraining the photoionization models with the lyman series transitions , and thus determines @xmath89 ( an example of this process is given in appendix [ app : constraints ] ) . by exploring a large range of @xmath82 and [ @xmath91/fe ]
, we verified the above relationships .
when only an upper limit is available on @xmath86 for a given cloud , one has two arbitrarily chosen parameters , [ @xmath91/fe ] and @xmath86 , which together uniquely determine the ionization parameter .
the constraints on the cloud ionization parameters , abundance pattern , and metallicities are fairly tight and robust ; even when @xmath86 was an upper limit , the , , and ratios were key for constraining the ionization parameter , independent of the abundance pattern ( see appendix [ app : constraints ] ) . in a given system , for clouds in which @xmath86 was only an upper limit
, we assumed that they were identical vis vis their column densities ( had identical metallicities and abundance patterns ) .
this yielded model clouds with identical @xmath82 , and [ @xmath91/fe ] , but unique @xmath80 , due to their unique @xmath87 .
the allowed range of cloud to cloud variations within a system , if desired , can be obtained from the two relations giving @xmath89 and @xmath94 ( as long as the total @xmath80 is held constant ) . to narrow parameter space
, we began with a grid of photoionization models with cloudy , where the grid was defined for ( 1 ) @xmath95 from @xmath96 to @xmath97 in intervals of 0.5 dex ( this provides the ionization parameter , @xmath83 ) , ( 2 ) @xmath80 from @xmath98 to @xmath99 in 1 dex intervals , ( 3 ) @xmath82 , from @xmath100 to @xmath101 in intervals of 0.2 dex , and ( 4 ) solar and @xmath93 = + 0.5 $ ] abundance pattern . once the parameter space was narrowed , we ran cloudy in its optimized mode tuned to the column densities , to obtain the adopted models .
model clouds are drawn from the grid and a synthetic fos spectrum is generated from the model column densities for the transitions listed in table [ tab : fosids ] ( end of paper ) .
the simulated fos spectrum is generated by modeling the absorption from each transition by voigt profiles convolved with the fos instrumental spread function .
the doppler parameters of the modeled transitions are determined from the observed @xmath46 parameters and the kinetic temperature , @xmath102 , output by the cloudy models , typically 5000 to 30,000 k. the cloudy temperature is used to estimate the thermal component of @xmath103 , from which the turbulent @xmath46 parameter is computed from the relation , @xmath104 .
the total @xmath46 parameter of any transition can then be estimated from its cloudy thermal @xmath46 and the turbulent @xmath46 . the synthetic spectrum
is then superimposed on the observed fos spectrum and the @xmath47 is calculated pixel by pixel in regions of interest as an indicator of the goodness of the model . as will be shown ,
the model clouds could not account for even a small fraction of the absorption strengths of the higher ionization species ( i.e. , s0.1emvi , , and ) .
thus , we postulated a high ionization diffuse component not seen in absorption .
a maximal diffuse scenario provides the low ionization limit of the clouds such that their contribution to the moderate and high ionization species is minimized .
a minimal diffuse scenario provides the high ionization limit of the clouds such that their contribution to the moderate and high ionization species is maximized . for a postulated high ionization component
, we used and to constrain model cloud properties for both a solar abundance pattern , @xmath105 } = 0 $ ] , and an oxygen to carbon enhancement of @xmath105 } = -0.5 $ ] . for constraints on the models from s0.1emvi , we use @xmath106 = 0 $ ] , but note that sulfur is enhanced by @xmath107 dex for @xmath108 < 0 $ ] .
we avoid using as a primary constraint because the chemical enrichment processes for nitrogen can lead to wide variations in its abundance ( @xcite ) .
we adopt @xmath109 } = 0 $ ] for this work , but occasionally discuss possible variations in this ratio .
to examine the range of properties of a diffuse component , we did the following systematic explorations .
for each system , we first assumed that any required high ionization gas is in a _ single _
, broad component .
the adopted models of the high ionization components are obtained using the same techniques illustrated in appendix [ app : constraints ] , but the primary constraints are the residual strengths of the and absorption unaccounted by the clouds . also , consistency with any unaccounted , , , s0.1emvi and is required .
we then explored the possibility that the arises in a few `` only '' clouds while the arises in a diffuse low density high ionization phase that is separate from the clouds , which are separate from the clouds .
the and s0.1emvi absorption profiles were critical for constraining these `` cloud '' models .
we systematically explored if contributions from both photoionization and collisional ionization are consistent with the data . for collisional ionization ,
we have drawn from the equilibrium models of sutherland & dopita ( 1993 ) , with solar abundances . in these models ,
the relative column densities of various species , including that of neutral hydrogen , are a unique function of temperature for a given metallicity .
thus , under the model assumptions , the remaining free parameter is the temperature .
to do so , we located the lowest ionization level of a photoionized diffuse component consistent with the intermediate and moderately high ionization species , after taking into account the clouds . a collisional ionization model was then tuned to any unaccounted absorption in and .
our intent is to present a general picture of both the cloud and high ionization diffuse component properties within the context of our modeling .
the adopted model properties are presented in tables [ tab : cloudpars ] , [ tab : pics ] , and [ tab : dics ] . in table
[ tab : cloudpars ] , we list the physical parameters of the twelve clouds for the scenarios of a maximal or minimal diffuse component . typical model parameters for each system are given in table [ tab : pics ] .
the tabulated values serve as a guide ; in the following discussion we quote allowed ranges in the cloud properties based upon [ sec : models ] .
the diffuse component properties are listed in table [ tab : dics ] .
the results described below are for the haardt & madau ( 1996 ) extragalactic background . in appendix
[ app : models ] , we describe our explorations with several galactic / starburst radiation fields ( representing both somewhat extreme and `` normal '' cases ) and argue that our general conclusions are not altered . in figure 3 , we present the synthetic spectrum of our models superimposed upon the fos spectrum .
the model transitions are labeled with three
point ticks , which give the locations of systems a , b and c , from blue to red , respectively .
three synthetic spectra are shown .
the dotted line spectrum is of the twelve photoionized clouds .
the thin solid line spectrum includes both the clouds and the single phase photoionized diffuse components .
the thick solid line spectrum includes a two phase photo plus collisionally ionized component in system a. since @xmath86 was not measured in any of the system a clouds , the clouds were assumed identical vis vis their @xmath87 .
the to ratio tightly constrains the ionization level of these six clouds .
there is only a very small range of allowed ionization conditions , with @xmath111 . to avoid super solar metallicity , the mean @xmath112 $ ] ranges from @xmath113 to @xmath114 , which corresponds to @xmath115 , respectively , for the best match to the lyman series .
we obtained @xmath116 in each of the six clouds ; system a makes a negligible contribution to the lyman limit break .
a highly ionized phase , not seen in , is required to account for the observed , , and absorption . for a single component , a doppler width of @xmath117 is consistent with the data .
we ran a series of simulations , focused on the blended profile , to explore the best combination of doppler widths for and in systems a and b , and to determine the allowed ranges for fits with with @xmath118 .
based upon these simulations , we adopted @xmath119 .
this component is consistent with either a single phase photoionized diffuse medium with @xmath120 , or a two phase photo plus collisionally ionized diffuse medium with @xmath121 and @xmath122 k , respectively .
all phases must be metal rich , @xmath123 , in order to not overproduce the lyman series .
it is not possible to establish which of the two provides a better description of the high ionization transitions .
the latter yields a better match to the absorption .
however , @xmath124 \simeq + 0.15 $ ] in the photoionized only phase could alternatively explain the absorption .
if the collisional component is present , further tests showed that arising in a few clouds with smaller @xmath46 parameters was not ruled out .
the bottom line is that a highly ionized phase is required , whether the arises in a single or multiple clouds and the and in a collisionally ionized phase , or whether the , , and all arise in a single highly photoionized phase .
for the photoionization scenario , the diffuse cloud size increased as @xmath46 was decreased from @xmath125 , primarily due to the curve of growth behavior of ; its column density increases rapidly with decreasing @xmath46 and is effectively constant for @xmath126 in the relevant range of equivalent width . based upon exploration of the allowed range of @xmath46 parameters , an upper limit on the size of the diffuse component is @xmath127 kpc for @xmath128 .
however , the best models , incorporating both a @xmath129 fit to the profile and matching to the , s0.1emvi , and profiles , had @xmath130 , which yielded sizes of @xmath31@xmath9 kpc .
a range of ionization conditions ( @xmath132 ) for the clouds are found for this system , primarily based upon the observed to ratios in clouds 8 , 9 , and 11 , and upon , and the other low ionization species for clouds 7 and 10 .
however , the low ionization limits ( maximal diffuse scenario ) are ruled out because the , , and a large fraction of the must arise in the clouds .
the lyman limit break is primarily produced by clouds 8 and 11 .
for these clouds , the metallicity must be @xmath133 , even for an @xmath91group enhanced abundance pattern , or else the lyman limit break would be too large . in clouds 7 and 10 ,
the allowed abundance pattern range is @xmath134 \leq + 0.5 $ ] , corresponding to @xmath135 ( the lower limit is constrained by the lyman series ) .
the arises primarily in cloud 10 , the most highly ionized , lowest density , and extended ( @xmath7 kpc ) cloud of the five . even for the minimal diffuse scenario ( high ionization limit ) ,
the predicted absorption is still well below the observed strength .
thus , an additional high ionization component , not seen in absorption , is required to account for the , , and absorption .
this component is consistent with a single phase photoionized diffuse medium ( @xmath136 , based upon the simulations of the profile ) with a solar abundance pattern and near solar metallicity .
the bulk of the , , and absorption arises in this component , whereas the and absorption arise primarily in cloud 10 .
the arises in both the clouds and the diffuse component . in figure 3@xmath66 , we point out the @xmath137 absorption , which accounts for the reduced flux in the lyman break .
we also point out the weak s0.1emvi @xmath78 absorption , which arises in the diffuse component ( the @xmath138 transition is blended with from system c ) .
it is not possible for a collisionally ionized phase to substantially contribute . as with system a
, we obtained a maximum line of sight size , @xmath139 , of @xmath140 kpc for the highly ionized diffuse component , but found a preferred size range of @xmath8@xmath9 kpc . for @xmath141
, the larger cloud size elevated absorption in all high ionization species such that , in particular , s0.1emvi was significantly overproduced .
a reduced @xmath106 $ ] would yield reduced sulfur absorption , but is unlikely because sulfur is usually enhanced relative to solar and is known to not suffer dust depletion ( @xcite ) . though system c is a single weak cloud , it is best described by two photoionized phases .
this inference is based upon a self consistent match to the lyman series , which is obtained when both a narrow and a broad component are included .
the narrower low ionization cloud accounts for the in the , , and absorption , whereas the broader high ionization diffuse component contributes significantly to the and profiles .
the lower ionization phase produces the narrow , , and in a smaller cloud .
we obtained @xmath143 and the range @xmath144 for @xmath145 \leq + 0.5 $ ] .
the higher ionization phase , with @xmath146 , has @xmath147 , which yields a size @xmath148 kpc .
it is not possible for this high ionization component to be strictly collisionally ionized nor is it possible for it to be a two phase photo plus collisionally ionized component ; any contribution from a collisionally ionized phase is insignificant .
as shown in figure [ fig : vpfits ] , system a is comprised of six distinct clouds with a total velocity spread of @xmath149 and is @xmath150 from system b , which has five clouds spread over @xmath6 .
system c is @xmath19 from system b and is comprised of a single , resolved cloud . in figure 3 , we show the normalized fos spectrum with simulated spectra superimposed .
these low resolution data reveal that each system is rich in multiple chemical species covering a wide range of ionization potentials and that the chemical and ionization conditions differ from system to system .
one motivation for our study was simply to ascertain if a multiphase medium was required to explain the strong , and absorption lines .
in all three systems , we were required to postulate a higher ionization component that is not seen in absorption .
we emphasize that ( 1 ) the overall cloud properties are well constrained by the data and modeling , and ( 2 ) the allowed cloud to cloud variations are constrained tightly enough that no scenario even remotely modifies the requirement for a
high ionization phase to explain the data . to the accuracy afforded by the fos spectrum ,
each of these high ionization phases is well described by a single component with @xmath151 ( based upon simulations ) .
given the large @xmath46 parameters and the range of sizes derived from the models ( table [ tab : dics ] ) , this high ionization gas is likely to have a line of sight extent of @xmath152 kpc , and thus may be a surrounding medium in which the @xmath153 kpc clouds are embedded . consider the cloud to cloud variations in system b , which is a lyman limit system .
the line of sight velocity spread of the clouds is @xmath154 , which implies they are bound within a galactic potential .
if the clouds are equally illuminated by the extragalactic background , then the presence of in three of the clouds ( 8 , 9 , and 11 ) implies that they are more dense , more shielded from the ionizing flux , and/or iron group enriched relative to the other two clouds ( 7 and 10 ) .
this suggests that clouds 8 , 9 , and 11 may be spatially contiguous ( relatively speaking ) , in that they may share similar histories of iron group enrichment from type ia sne . in the galaxy
, the association of type ia explosions with the kinematically old disk implies that some events take place at large scale heights , so there is uncertainty in how much iron
rich gas is driven into galactic halos ( @xcite ) .
the unique cloud in system b is cloud 10 , which gives rise to a broader absorption profile ( @xmath155 ) and has no detectable .
it has a higher ionization condition and gives rise to the majority of the absorption ( see figure [ fig : stis ] ) .
it also has the largest @xmath87 .
models yield that it is extended ( @xmath7 kpc ) and has lower metallicity .
this leads us to conservatively speculate that cloud 10 is more akin to a halo like cloud or to a so called galactic high velocity cloud .
the ratio @xmath156 may be a useful indicator of the differing local environments of clouds in higher redshift systems .
system c classifies as a `` weak '' absorbers , defined by @xmath20 ( @xcite ) . from a sample of thirty such systems over the redshift range @xmath157 , churchill et al
. found a wide range of @xmath158 and @xmath159 , presumedly due to variations in abundance pattern and ionization conditions , including single phase and multiphase .
these absorbers are sub
lyman limit systems with @xmath160 and [ some with @xmath161 ( @xcite ) ] .
apart from its line of sight proximity ( @xmath162 ) to system b , system a would classify as a weak absorber .
the @xmath163 kinematic spread of the six clouds in system a , and the large @xmath164 ratio in the diffuse component are suggestive of lower ionization clouds moving within a high ionization galactic corona , or halo , where the ratio @xmath164 is expected to be large ( @xcite ) .
is it possible that the system c cloud arises in a similar environment as the six system a clouds , but that the line of sight happens to sample only one cloud ?
if so , the absorption strength in system c would be comparable to that of system a , and it is significantly weaker .
for weak systems , it may be that strong , broad absorption implies a larger number of clouds with a larger kinematic spread . in figure
[ fig : stis ] , we present simulated stis/_hst _ spectra with @xmath165 , two pixels per resolution element , and a signal to noise ratio of @xmath140 for the first four transitions in the lyman series , and for the , , , and profiles of systems a and b. these spectra were generated assuming voigt profiles with the properties listed in tables [ tab : pics ] and [ tab : dics ] .
these model profiles can be compared directly to observed data ( from stis/_hst _ ) and thus provide a direct test of the models .
we show the contributing components as smooth curves .
the dotted line curves are the clouds , with their velocity centroids marked with the short ticks above the continuum .
the solid curves are from the photoionized diffuse component and the dash dot curves in system a represents the collisionally ionized diffuse component . here
, we emphasize the importance of the lyman series .
for all three systems , the and profiles in the fos spectra were significantly broader than could be fully accounted by the obtained soley from the model clouds .
however , the narrower , , and profiles were fully accounted by the in these clouds . based upon the curve of growth behavior of the lyman series , we found that the addition of a very broad ( @xmath166 ) , lower @xmath80 component naturally explained the deeper and broader and profiles , without overproducing the narrower , , and profiles ( or modifying the lyman limit break ) .
such behavior of the lyman series in low resolution data is likely to be a strong indication that a broad , perhaps highly ionized diffuse component is present . in figure
[ fig : stis ] , we illustrate this behavior as it would be seen in higher resolution spectra .
note that the widths of and are dominated by a broad , high ionization component , whereas the higher order transition widths are dominated by the clouds . in high resolution spectra ,
a diffuse component gives rise to a broadened , shallow wing when the profiles are saturated ( see and in system b ) and/or suppresses the recovery of the flux to the continuum level between clouds in the profile centers ( see and in system a ) . in system a , the profiles have structure due to the lower ionization clouds .
the absorption predominantly arises in the clouds , and thus closely traces the kinematics .
together , these profiles are tantalizing similar to those observed by savage , sembach , & cardelli ( 1994 ) along the line of sight to hd 167756 in the galaxy .
we quote , `` the sight line contains at least two types of highly ionized gas .
one type gives rise to a broad profile , and the other results in a more structured profile .
the profile contains contributions from both types of highly ionized gas .
'' we also find similarities between the model profiles and those in the damped systems at @xmath167 toward q@xmath168 and @xmath169 toward q@xmath170 ( lu et al .
1996 , figures 2 and 3 ) . in system b ,
cloud 10 is clearly unique among the clouds .
this cloud has the largest ionization parameter and accounts for the majority of the absorption , which was clearly constrained to arise in the clouds .
since the ionization conditions of clouds 8 , 9 , and 11 were set by their measured @xmath86 , the ionization condition in cloud 10 was well constrained by the ratio @xmath171 ; there was no alternative but for cloud 10 to dominate the absorption .
the point is that this cloud may arise in a spatially distinct environment and have a unique formation history from the other clouds in system b ( see [ sec : discussion ] ) .
the profile is dominated by a diffuse higher ionization phase , whereas the is predominantly due to higher density clouds .
some lines of sight through the galaxy also exhibit narrow profiles and broad profiles ( for examples , see @xcite ; @xcite ) . in contrast , the and profiles in the @xmath172 damped system towards q@xmath173 ( @xcite ) appear to arise in the same phase .
a synthetic stis/_hst _
spectrum for system c ( not shown ) reveals a narrow profile and broad , , and profiles .
the profile exhibits a slightly deep , narrow core due to the low ionization phase .
however this contribution would be lost for the expected noise levels in observed data , unless this narrow component was off center by @xmath174 relative to the broad component ( within the uncertainty of our modeling such an offset is not ruled out ) .
within 100 of the qso , kirhakos et al .
( 1994 ) identified @xmath8 galaxies with @xmath175 .
compared to field galaxies , this is a slight overdensity by a factor of a few to several ( @xcite ) . identified within 10 of the qso are three bright galaxies with impact parameters @xmath27 , @xmath28 , and @xmath29 kpc , respectively ( @xmath30 ) . at @xmath176 , thimm ( 1995 )
detected a @xmath177 flux of @xmath24 ergs s@xmath25 from the galaxy at @xmath178 kpc .
given these facts , we entertain the possibility that the absorption systems arise in individual galaxies either in a small group or possibly in a cluster .
such a possibility would have interesting implications in view of the types of mechanisms that could give rise to highly ionized gas in such environments .
mulchaey et al .
( 1996 ) predicted that poor groups that are rich in spiral galaxies have hot , diffuse coronae that are cooler than the x ray coronae surrounding poor groups dominated by e / s0 galaxies .
this high ionization intragroup material is predicted to have @xmath179 k and to primarily give rise to strong , whereas any or absorption is predicted to arise in the proximity of the galaxies themselves ( @xcite ) . as such
, the profiles would be broader than the and .
the same would hold for the lyman series ( @xcite ) .
our models can fully explain the , , and in a single component of @xmath180 k gas ( the possible collisional component in system a has @xmath181 k ) . the observed or widths
are not broader than those of the and , and their @xmath70 @xmath46 parameters are not suggestive of gas that is kinematically akin to the @xmath163 dispersion of a galaxy group .
the , , and profiles are clearly aligned with the three absorption redshifts , suggesting that the high ionization material is spatially coincident with the individual systems . the upper limit on the size of the photoionized diffuse components is @xmath182 kpc ( assuming photoionization equilibrium ) , and the inferred @xmath123 metallicities suggest that the high ionization material surrounds the galaxies and has been enriched by them .
significantly lower metallicities are expected if the gas was intragroup material left over from the formation processes .
we do not favor an interpretation of the data in which the high ionization material arises in a intragroup or common halo .
the highly ionized material in these three @xmath10 galaxies may be more akin to galactic coronae ( see spitzer 1956 , 1990 ; savage et al .
1997 ) , material stirred up by energetic mechanical processes , such as galactic fountains . in this scenario ,
the gas is concentrated around individual galaxies which presumably provide a source of support , heating , and chemical enrichment . in the galaxy ,
@xmath183 was measured for , and was observed to slightly increase with the ionization level of the transition ( @xcite ) .
simulations of the allowed range of @xmath46 parameters for systems a and b yielded similar doppler widths , with @xmath136 providing a good match to the , , and profiles for both systems . if the high ionization diffuse components in these systems are arising in coronae analogous to that surrounding the galaxy
, they may have similar turbulent processes supporting their scale heights .
the radial extent of the galactic corona is unknown , but for local galaxies , radio maps reveal extends to tens of kpc ( @xcite ; @xcite ) , beyond which the hydrogen becomes optically thin and a highly ionized extension is expected ( @xcite ; @xcite ; @xcite ) .
the impact parameters of the candidate galaxies are @xmath184@xmath185 kpc , with the orientation of the line of sight through each galaxy unknown .
absorption , even from non spherical absorbers , is not unexpected at an impact parameter of @xmath37 kpc ( see figure 1 of @xcite ) . for the galaxy , savage et al .
( 1997 ) measured the effective scale heights of @xmath186 , @xmath187 , and @xmath188 kpc for , , and , respectively .
we find a preferred size of @xmath8@xmath9 kpc for the size of the high ionization gas at @xmath189 , with an upper limit of @xmath127 kpc . since these `` sizes '' ( or the path length through an oriented structure ) are comparable to twice the galactic scale height , the distribution of high ionization is consistent with galactic like coronae .
a scenario in which the high ionization gas arises in individual galactic coronae is in contrast to that proposed by lopez et al .
( 1998 ) for a system at @xmath190 in the spectra of he @xmath191 a , b. in a double line of sight study , lopez et al . found profiles consistent with @xmath192 and that the extent of the highly ionized gas was @xmath6 kpc .
these inferences also differ from those of bergeron et al .
( 1994 ) , who found that the phase in the @xmath193 absorber toward pks @xmath194 was at least @xmath195 kpc in extent .
we have studied the kinematic , chemical , and ionization conditions of three metal line absorption systems at @xmath10 seen in the pg @xmath5 spectrum .
the systems were selected by the presence of absorption in a high resolution spectrum , and were chosen as a pilot study of a larger program designed to chart the physical conditions and evolution of absorbing gas in galaxies .
the profiles are shown in figure [ fig : order28 ] , with each system designated a , b , and c. rich absorption line data from fos/_hst _ ( @xcite ) revealed strong , , and absorption , as well as and many other species and transitions covering a wide range of ionization potentials .
based imaging data ( @xcite ) revealed three candidate absorbing galaxies within 10 of the qso and possibly a group of galaxies within 100 of the qso .
the main goals of the study were to see if multiphase gas was required to explain the strong high ionization absorption line data , and to infer some level of information on the gas metallicities and spatial distributions . assuming the clouds are photoionized
, we found the range of chemical and ionization conditions consistent with both the high resolution and low resolution data .
we then postulated the presence of high ionization components not seen in absorption to explain the unaccounted , , , and absorption .
we briefly summarize the main results of our study : \1 .
for systems a and b , we were required to postulate a high ionization phase in addition to the lower ionization clouds .
system c could be made marginally consistent with a single phase absorber , though a two phase absorber is strongly preferred due to the nature of the lyman series absorption .
we infer that , in these systems , the lower ionization clouds arise in high ionization diffuse gas ; each of these absorption systems is comprised of a multiphase gaseous medium .
we find that the high ionization phase of system a _ could _ arise in multiple , narrower `` clouds '' .
for system b , such clouds are ruled out .
the absorbing gas in both the clouds and the high ionization components are consistent with photoionized clouds . in systems
b and c , a collisionally ionized phase is ruled out . only in system
a could the data be made consistent with a three phase absorber , incorporating the photoionized clouds , a highly photoionized diffuse component , and a collisionally ionized component .
this three phase model provided a more consistent match to the absorption in this system . in this three phase scenario ,
the could arise in a few narrower components .
however , @xmath124 \simeq 0.15 $ ] in the highly photoionized component , instead of the assumed @xmath124 = 0 $ ] , would remove the need for the collisionally ionized gas and rule out the three
phase absorber .
we find no evidence that the gas is in a separate and very highly ionized diffuse phase that encompasses the and absorption .
based upon the @xmath196 profile widths , inferred @xmath123 metallicities , @xmath197 k temperatures , inferred @xmath198 kpc sizes , and the clear redshift alignment of the high ionization transitions with the systems , we suggest that the high ionization gas is analogous to the galactic corona in that it traces the galaxies themselves and does not appear to exhibit the characteristics predicted for intragroup or intracluster material .
the likely arises in the same phase as the and .
the is constrained to arise in the same phase as the clouds , whereas arises in both the clouds and the high ionization phase .
we have found cloud to cloud variations in the chemical and ionization conditions in the five clouds of system b. three of the clouds are likely to be iron
group enriched and have higher densities and low ionization conditions .
the majority of the neutral hydrogen giving rise to the lyman break in the fos spectrum is from these clouds .
a lower metallicity cloud giving rise to a broader absorption profile is interspersed in velocity with these clouds , likely has an @xmath91group enhanced abundance pattern , and gives rise to the majority of the absorption .
we speculate that this cloud may be similar to a halo like cloud , and suggest that the ratio @xmath199 may be a useful indicator for discriminating between clouds in different parts of high redshift galaxies .
the most compelling reason why we favor the scenario in which the high ionization diffuse material is coupled to the galaxies is the clear kinematic separation of the profiles corresponding to systems a and b. each of the inferred diffuse components must be centered ( at least roughly ) on the systems , and they must be distinct from one another in velocity space .
there are additional , if less compelling , arguments .
the inferred @xmath46 parameters in the model diffuse components are in the same regime as those found for the galactic corona ( @xcite ) , further suggesting that the material is galaxy associated .
the inferred metallicities are high , @xmath200 , which is best understood if the material had been enriched by its host galaxy and further suggests that the origin and source of enrichment of the diffuse gas is related to star forming parts of galaxies .
the enrichment could be due to _ in situ _ star formation as gas clouds collide and cool in the galactic halos ( @xcite ) , or could be due to galactic fountain processes from the galactic disks ( @xcite ) . as a speculative aside
, we ask if galactic
like coronae absorbers are likely to be a common form of systems at @xmath201 , as opposed to group halos .
burles & tytler ( 1996 ) have shown that absorbers selected by the presence of have the same redshift path density as lyman limit absorbers at @xmath202 . as we have found here , can be associated with a lyman limit system when the absorbing gas is segregated into multiple ionization phases . if multiphase absorption is common in absorbers , some might arise in galactic like corona .
it would seem that this pilot study has shown that wholesale study of the kinematic , chemical , and ionization conditions of absorbers , using high resolution profiles and the available low resolution _ hst _
spectra , would yield a improved understanding of galactic gas at early epochs . in order to assess the robustness of our modeling ,
we have synthesized high resolution stis/_hst _ spectra of the ultraviolet transitions ( presented in figure [ fig : stis ] ) .
the modeling techniques applied in this paper can be directly tested by comparing these predicted profiles with those observed with stis/_hst_. if our approach proves to yield an accurate description of the gas , then wholesale modeling can be embarked upon for roughly 50 systems without requiring large amounts of space based telescope time to acquire high resolution spectra of high quality .
this work was supported by nsf ast9529242 and ast9617185 and by nasa nag56399 and ar07983.0196a from stsci .
special thanks to buell jannuzi , sofia kirhakos , and don schneider for generously supplying the fos/_hst _ spectra and for helpful discussion regarding the reduction and analysis of the fos data .
we are thankful for comments from the anonymous referee that resulted in an improved manuscript .
we thank karen knierman and jane rigby for assistance in producing the stacked spectrum of the system a clouds for the limit on @xmath203 , and are grateful to steven s. vogt for the hires spectrograph .
in this appendix , we demonstrate our modeling methodology for constraining the ionization parameters , metallicities , and overall cloud properties using the fos/_hst _ data . in figure [
fig : siii - siiv ] , we present an example of how the data are used to constrain the ionization parameter , @xmath83 . in figure
[ fig : siii - siiv]@xmath66 the synthetic fos spectrum of the si0.1emii @xmath61 doublet is superimposed on the fos data for the @xmath50 upper limits on @xmath204 , and for the set ratios @xmath205 , @xmath206 , and @xmath207 . the si0.1emiv @xmath59 doublet is shown in figure [ fig : siii - siiv]@xmath46 .
the assumed abundance pattern is solar .
clouds 8 , 9 , and 11 have their @xmath83 constrained by the observed @xmath208 . the ratio of to can uniquely determine @xmath83 for the remaining clouds . for system
a , the best match is provided by @xmath209 [ which corresponds to @xmath210 .
lower ionization clouds are not possible because they overproduce , whereas higher ionization clouds overproduce . for system
b , a reasonable match to the observed and is achieved if clouds 7 and 10 ( dominated by 11 because of its larger column density ) are assigned @xmath211 , which corresponds to @xmath212 .
less ionized clouds are possible , but an additional more highly ionized component would then be required to produce . in figure [
fig : hi - z ] , we present an example of how the lyman series and limit are used to constrain the metallicity for the ionization parameters determined in the above illustration
. we have assumed @xmath213 } = 0 $ ] for this illustration .
three metallicities are shown , @xmath214 , @xmath215 , and @xmath101 , to illustrate the strength variations in the lyman series and limit . at low metallicity
, a large @xmath80 is needed to produce the observed metal lines with low and intermediate ionization levels .
such a large @xmath80 can be inconsistent with the observed lyman series lines and lyman limit break . as described in
[ sec : data ] , the lyman limit break implies a total @xmath80 of @xmath216 . for system
a , the best match to the lyman series lines is given by @xmath217 , super solar metallicity . for system
b , the best match also has high metallicity ; clouds 8 , 9 , and 11 must have @xmath218 , and cloud 10 must also have near solar metallicity in order that there is not too large an @xmath80 .
it is important to point out that the assumed abundance pattern directly affects the inferred metallicities .
if the abundance pattern is @xmath91group enhanced , @xmath213 } = + 0.5 $ ] , the illustrated metallicities would proportionally drop by @xmath107 dex .
this is because the cloud properties are tuned to the @xmath87 , and magnesium is an @xmath91group element ( see [ sec : models ] ) .
given the @xmath23 emission measured by thimm ( 1995 ) , it is reasonable to assume that high energy photons could be escaping the galaxies ( e.g. @xcite ) .
the thimm measurement of an flux from one galaxy ( and limits for the others ) constrains the contribution of galactic ionizing photons to be @xmath219 s@xmath25 under the assumption that 50% of the photons escape from the galaxy .
this is a factor of five to 50 times greater than that estimated for the galaxy ( @xcite ) and for external galaxies ( @xcite ) , respectively . keeping this generous upper limit in mind
, we consider the effect a change in spectral shape could have on our main results .
several contrasting starburst models from bruzual & charlot ( 1993 ) were explored . in all cases ,
the galaxy spectra were normalized relative to the haardt & madau ( 1996 ) flux , @xmath220 , at 1 rydberg .
depending on the spectral shape , the above limit corresponds to @xmath221@xmath8 , where @xmath222 is the flux of the starburst galaxy model at 1 rydberg .
first we consider a @xmath223 year instantaneous burst model , because of its prominent edge [ see fig . 4 of bruzual & charlot ( 1993 ) ] .
as the ratio @xmath224 is increased , the hydrogen becomes more ionized and it is possible to decrease the cloud metallicities without overproducing .
however , only a modest decrease of 0.4 dex is possible for @xmath225 .
for these models , there are relatively fewer photons capable of ionizing and thus there is even less produced in the clouds .
similarly , in order to maintain the observed to ratio , it is necessary to slightly increase the cloud ionization parameter by a few tenths of a dex from the pure haardt & madau case .
we also consider the effect of the change in spectral shape upon the diffuse phase that produces and .
similar conclusions hold for the metallicity , which can decrease by at most 0.5 dex from the pure haardt & madau case ; the diffuse phase still must have relatively high metallicity ( @xmath226 ) . for energies greater than the edge ,
this overall spectrum is unchanged relative to haardt & madau , and therefore the balance between and , and other high ionization energy transitions is unaltered . in order to explore a severe edge
, we explored another extreme model , a @xmath227 year instantaneous burst . in this case
the measured constraint also forces the upper limit @xmath225 .
interestingly , the cloud metallicity can not be reduced ( relative to pure haardt & madau ) because @xmath80 actually increases relative to @xmath87 as the galaxy contribution is increased . due to similar behavior
, the metallicity of the required diffuse component is constrained to be solar or super solar .
the change in spectral shape has the effect of increasing and s0.1emvi , but decreasing .
the bulk of the still arises in the clouds , but s0.1emvi in the diffuse phase is overproduced , and the required super solar metallicity seems implausible . finally , we consider a later type galaxy model with a 16 gyr stellar population and with exponentially decreasing but persistent , star formation [ the @xmath228 model ( @xcite ) ] .
this model has 1% of the total star forming mass in stars after 1 gyr .
the star formation longevity of this relatively quiescent galaxy is more plausible than an instantaneous burst . as the galactic flux
is incrementally increased relative to the extragalactic background , the metallicity constraints are unchanged , but the ratio @xmath229 decreases due to the drop in flux at the edge .
thus , the requirement for a diffuse component in these absorbers is strengthened if galactic flux is contributing .
we also found that it is increasingly difficult to produce the required @xmath230 ratio as the galactic flux is increased unless the ionization parameter is significantly increased .
this results in the destruction of .
thus , our conclusion that the must arise in the clouds is also strengthened if such a galactic flux is contributing to or dominating the ionizing spectrum .
likewise , our conclusion holds that any collisionally ionized diffuse component in system b must be negligible . as with the extragalactic background scenario ,
the becomes too large and inconsistent with the data if one lowers the ionization condition in the photoionized diffuse component .
although some starburst galaxy spectra can somewhat decrease the constraint on the cloud and diffuse component metallicities , this effect is limited to less than 0.5 dex due to constraints on emission .
these galaxies are not strong starbursts , though they could have modest galaxy contribution to the spectral shape .
thus , we find that our _ general _ conclusions with regard to the requirement of highly ionized diffuse components and relatively high metallicity clouds are not sensitive to the assumed spectral shape of the ionizing flux , though the details of the adopted models would be somewhat modified .
churchill , c. w. 1997a , in proceedings of the 13th iap colloquium : structure and evolution of the igm from qso absorption line systems , ed .
p. petitjean & s. charlot , ( paris : editions frontires ) , 229 churchill , c. w. 1997b , in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution , eds .
w. h. waller , m. n. fanelli , j. e. hollis , & a. c. danks ( new york : aip press ) , 313 crcrccccccr a & 1 & 0.92501 & @xmath231 & & @xmath232 & @xmath233 & & @xmath234 & @xmath235 & @xmath236 + a & 2 & 0.92535 & @xmath237 & & @xmath238 & @xmath239 & & @xmath240 & @xmath235 & @xmath241 + a & 3 & 0.92550 & @xmath242 & & @xmath243 & @xmath244 & & @xmath240 & @xmath235 & @xmath245 + a & 4 & 0.92595 & @xmath246 & & @xmath247 & @xmath248 & & @xmath240 & @xmath235 & @xmath249 + a & 5 & 0.92639 & @xmath250 & & @xmath251 & @xmath252 & & @xmath253 & @xmath235 & @xmath236 + a & 6 & 0.92648 & @xmath254 & & @xmath255 & @xmath256 & & @xmath253 & @xmath235 & @xmath257 + b & 7 & 0.92720 & @xmath258 & & @xmath259 & @xmath260 & & @xmath234 & @xmath235 & @xmath261 + b & 8 & 0.92742 & @xmath262 & & @xmath263 & @xmath264 & & @xmath265 & @xmath266 & @xmath267 + b & 9 & 0.92764 & @xmath268 & & @xmath269 & @xmath270 & & @xmath271 & @xmath272 & @xmath273 + b & 10 & 0.92780 & @xmath274 & & @xmath275 & @xmath276 & & @xmath277 & @xmath235 & @xmath278 + b & 11 & 0.92803 & @xmath279 & & @xmath280 & @xmath281 & & @xmath282 & @xmath283 & @xmath284 + c & 12 & 0.93428 & @xmath285 & & @xmath286 & @xmath287 & & @xmath288 & @xmath235 & @xmath245 + [ tab : vpfits ] rrccrccl 1a & 1762.92 & & 915.61 & 29.6 & a & & + 2a & 1765.03 & & 915.61 & & b & & matches depression at lyman limit + 3a & 1770.98 & & 915.61 & & c & & + 4a & 1797.13 & s0.1emvi & 933.38 & 88.0 & a & & c - fit ?
+ 5a & 1799.19 & s0.1emvi & 933.38 & & b & & bl from sys c ; c fit ? ; note + 6a & 1805.42 & s0.1emvi & 933.38 & & c & & bl7a ; c - fit ?
+ 7a & 1805.65 & & 937.80 & 13.6 & a & & bl6a ; c - fit ?
+ 8a & 1807.80 & & 937.80 & & b & & + 9a & 1813.90 & & 937.80 & & c & & + 10a & 1818.58 & s0.1emvi & 944.52 & 88.0 & a & & + 11a & 1820.66 & s0.1emvi & 944.52 & & b & & + 12a & 1826.97 & s0.1emvi & 944.52 & & c & & bl13a + 13a & 1828.64 & & 949.74 & 13.6 & a & & bl12a + 14a & 1830.82 & & 949.74 & & b & & + 15a & 1836.99 & & 949.74 & & c & & + 16b & 1872.52 & & 972.54 & & a & & + 17b & 1874.76 & & 972.54 & & b & & + 18b & 1881.08 & & 972.54 & & c & & bl19b ; bl + 19b & 1881.15 & & 977.02 & 47.9 & a & & bl18b + 20b & 1883.40 & & 977.02 & & b & & + 21b & 1889.75 & & 977.02 & & c & & + 22b & 1905.76 & & 989.80 & 47.4 & a & & bl or ? + 23b & 1908.04 & & 989.80 & & b & & + 24b & 1914.47 & & 989.80 & & c & & + 25c & 1974.93 & & 1025.72 & 13.6 & a & & + 26c & 1977.28 & & 1025.72 & & b & & + 27c & 1983.95 & & 1025.72 & & c & & + 28c & 1986.87 & & 1031.93 & 138.1 & a & & + 29c & 1989.25 & & 1031.93 & & b & & + 30c & 1995.36 & & 1036.33 & 24.3 & a & & bl31c ; bl ? + 31c & 1995.95 & & 1031.93 & 138.1 & c & & bl30c ; bl ? + 32c & 1997.75 & & 1036.34 & 24.3 & b & & bl33c + 33c & 1997.83 & & 1037.62 & 138.1 &
a & & bl32c + 34c & 2000.21 & & 1037.62 & & b & & bl ? + 35c & 2004.48 & & 1036.34 & 24.3 & c & & + 36c & 2006.96 & & 1037.62 & 138.1 & c & & + 37e & 2292.03 & & 1190.42 & 16.3 & a & & + 38e & 2294.76 & & 1190.42 & & b & & + 39e & 2297.56 & & 1193.29 & & a & & + 40e & 2300.31 & & 1193.29 & & b & & + 41e & 2302.50 & & 1190.42 & & c & & + 42e & 2308.06 & & 1193.29 & & c & & + 43d & 2323.00 & & 1206.50 & 33.5 & a & & + 44d & 2325.77 & & 1206.50 & & b & & + 45d & 2333.61 & & 1206.50 & & c & & + 46d & 2340.65 & & 1215.67 & 13.6 & a & & + 47d & 2343.45 & & 1215.67 & & b & & + 48d & 2351.35 & & 1215.67 & & c & & + 49d & 2385.23 & & 1238.82 & 97.9 & a & & bl galactic + 50d & 2388.08 & & 1238.82 & & b & & + 51d & 2392.89 & & 1242.80 & & a & & + 52d & 2395.75 & & 1242.80 & & b & & bl53d + 53d & 2396.13 & & 1238.82 & & c & & bl52d + 54d & 2403.83 & & 1242.80 & & c & & + 55f & 2426.82 & & 1260.42 & 16.3 & a & & + 56f & 2429.72 & & 1260.42 & & b & & bl ? + 57f & 2437.91 & & 1260.42 & & c & & + 58 g & 2569.51 & & 1334.53 & 24.4 & a & & + 59 g & 2572.58 & & 1334.53 & & b & & bl ? + 60 g & 2581.25 & & 1334.53 & & c & & bl ? + 61h & 2683.54 & & 1393.76 & 45.1 & a & & note + 62h & 2686.74 & & 1393.76 & & b & & + 63h & 2695.80 & & 1393.76 & & c & & + 64h & 2700.89 & & 1402.77 & & a & & + 65h & 2704.12 & & 1402.77 & & b & & + 66h & 2713.24 & & 1402.77 & & c & & + 67i & 2980.89 & & 1548.20 & 64.5 & a & & + 68i & 2984.46 & & 1548.20 & & b & & bl69i + 69i & 2985.85 & & 1550.77 & & a & & bl68i + 70i & 2989.42 & & 1550.77 & & b & & + 71i & 2994.52 & & 1548.20 & & c & & + 72i & 2999.50 & & 1550.77 & & c & & + [ tab : fosids ] lcccccccccl @xmath289 & & 0.92572 & & 0.92572 & 0.92572 & & 0.92768 & & 0.93428 & + @xmath82 & & 0 & & 0 & 0 & & 0 & & @xmath290 & + @xmath291 $ ] & & 0 & & 0 & 0 & & 0 & & 0 & + @xmath46 & & 70 & & 70 & 70 & & 70 & & 40 & km s@xmath25 + @xmath292 & & 11.3 & & 12.1 & 12.0 & & 12.8 & & 10.3 & @xmath293 + @xmath294 & & 13.3 & & 13.7 & 12.5 & & 14.3 & & 12.8 & @xmath293 + @xmath295 & & 13.8 & & 14.2 & 11.8 & & 14.8 & & 13.4 & @xmath293 + @xmath296 & & 14.5 & & 14.5 & 13.5 & & 15.2 & & 13.9 & @xmath293 + @xmath297 & & 12.7 & & 13.1 & 13.4 & & 13.7 & & 12.1 & @xmath293 + @xmath298 & & 14.2 & & 13.8 & 14.2 & & 14.5 & & 13.6 & @xmath293 + @xmath299 & & 15.0 & & 14.3 & 15.0 & & 14.9 & & 14.5 & @xmath293 + @xmath300 & & @xmath301 & & @xmath302 & & & @xmath302 & & @xmath303 & + @xmath80 & & 14.6 & & 14.8 & 13.5 & & 15.5 & & 14.5 & @xmath293 + @xmath304 & & 18.7 & & 18.5 & 19.2 & & 19.1 & & 18.6 & @xmath293 + @xmath305 & & @xmath207 & & @xmath306 & & & @xmath306 & & @xmath307 & @xmath308 + @xmath139 & & 16.5 & & 3.9 & @xmath309 & & 16.0 & & 11.6 & kpc + @xmath310 & & 1.9 & & 5.0 & 0.2 & & 5.0 & & 2.0 & + @xmath311 & & 0.3 & & 1.6 & 0.03 & & 1.6 & & 0.3 & + @xmath164 & & 1320 & & 250 & 30 & & 220 & & 4000 & + [ tab : dics ] | we used hires / keck profiles ( @xmath0 ) of and in combination with fos/_hst _ spectra ( @xmath1 ) to place constraints on the physical conditions ( metallicities , ionization conditions , and multiphase distribution ) of absorbing gas in three galaxies at @xmath2 , @xmath3 , and @xmath4 along the line of sight to pg @xmath5 .
the chemical and ionization species covered in the fos/_hst _ spectra are , , , , , , , , , , s0.1emvi , , and , with ionization potentials ranging from 13.6 to 138 ev .
the multiple clouds exhibit complex kinematics and the , and are exceptionally strong in absorption .
we assumed that the clouds are photoionized by the extra galactic background and determined the allowed ranges of their physical properties as constrained by the absorption strengths in the fos spectra .
a main result of this paper is that the low resolution spectra can provide meaningful _ constraints _ on the physical conditions of the clouds , including allowed ranges of cloud to cloud variations within a system .
we find that the clouds , which have a typical size of @xmath6 pc , give rise to the , the majority of which arises in a single , very large ( @xmath7 kpc ) , higher ionization cloud .
however , the clouds can not account for the strong , , and absorption .
we conclude that the clouds are embedded in extended ( @xmath8@xmath9 kpc ) , highly ionized gas that gives rise to , , and ; these are multiphase absorption systems .
the high ionization phases have near solar metallicity and are consistent with galactic like coronae surrounding the individual galaxies , as opposed to a very extended common `` halo '' encompassing all three galaxies .
= 100000 |
9 drees m and gerbier g 2006 _ j. phys . _ * g 33 * 233 , and references therein .
bottino a et al .
2005 _ phys .
* d 72 * 083521 he x g et al .
2007 _ mod .
lett . _ * a 22 * 2121 gondolo p and gelmini g 2005 _ phys . rev . _ * d 71 * 123520 damour t and l.m .
krauss 1998 _ phys .
lett . _ * 81 * 5726 ; collar j i 1999 _ phys .
* d 59 * 063514 gelmini g , kusenko a and nussinov s 2002 _ phys .
lett . _ * 89 * 101302 angloher g et al .
2002 _ astropart .
* 18 * 43 wong h t 2004 _ mod .
lett . _ * a 19 * 1207 wong h t et al . 2006 _ j. phys .
ser . _ * 39 * 266 wong h t and li h b 2005 _ mod . phys .
lett . _ * a 20 * 1103 li h b et al .
2003 _ phys .
lett . _ * 90 * 131802 ; wong h t et al .
2007 _ phys .
* d 75 * 012001 xin b et al . 2005 _ phys .
* d 72 * 012006 chang h m et al .
2007 _ phys .
* d 75 * 052004 barbeau p a , collar j i and tench o 2007 _ jcap _ * 09 * 009 these proceedings , for complete references and latest updates . | the report describes the research program on the development of ultra - low - energy germanium detectors , with emphasis on wimp dark matter searches .
a threshold of 100 ev is achieved with a 20 g detector array , providing a unique probe to the low - mas wimp .
present data at a surface laboratory is expected to give rise to comparable sensitivities with the existing limits at the @xmath0 wimp - mass range .
the projected parameter space to be probed with a full - scale , kilogram mass - range experiment is presented .
such a detector would also allow the studies of neutrino - nucleus coherent scattering and neutrino magnetic moments . as - texono/07 - 05 + weakly interacting massive particles ( wimp , denoted by @xmath1 ) are the leading candidates for cold dark matter ( cdm ) @xcite , and supersymmetric ( susy ) particles are the favored wimp candidates .
the popular susy models prefer wimp mass ( @xmath2 ) of the range of @xmath3100 gev , though light neutralinos remain a possibility @xcite . simple extensions of the standard model with a singlet scalar favors light wimps @xcite .
most cdm experiments optimize their design in the high - mass region , and have diminishing sensitivities for @xmath4 , where there is an allowed region if the dama annual modulation data are interpreted as wimp signatures @xcite . to probe this low - mass region , detectors with sub - kev threshold are necessary .
these low - threshold detectors will also open the window to look for wimps bound in the solar system @xcite , as well as non - pointlike susy candidates like q - balls @xcite .
sensitivity to sub - kev energy presents a formidable challenge to detector technology and to background control .
so far , only the cresst - i experiment have derived exclusion limits @xcite with sapphire(@xmath5)-based cryogenic detector at a threshold of 600 ev .
a research program in low energy neutrino and astroparticle physics is pursued @xcite by the texono collaboration at the kuo - sheng(ks ) reactor laboratory .
a scientific goal is to develop advanced detectors with kg - size target mass , 100 ev - range threshold and low - background specifications for wimp searches as well as for the studies of neutrino - nucleus coherent scatterings @xcite and neutrino magnetic moments ( @xmath6 ) @xcite .
the ks laboratory is located 28 m from a 2.9 gw reactor core and has an overburden of about 30 meter - water - equivalence ( mwe ) .
its facilities are described in ref .
@xcite , where the @xmath6-studies with a 1.06 kg germanium detector ( hpge ) at a hardware threshold of 5 kev were reported .
this hpge has also been used for the studies of reactor electron neutrons @xcite as well as searches for reactor axions @xcite .
the experimental procedures were well - established and the background above 12 kev were measured .
in particular , a background level of about @xmath7(cpd ) comparable to those of other underground cdm experiments was achieved .
`` ultra - low - energy '' germanium ( ulege ) detectors , developed originally for soft x - rays detection , are candidate technologies to meet the challenges of probing into the unexplored sub - kev energy domain @xcite .
these detectors typically have modular mass of 5 - 10 grams
. detector array of up to n=30 elements have been successfully built , while there are recent advances in developing single - element ulege of kg - size mass @xcite .
various prototypes based on this detector technology have been constructed .
depicted in figure [ fe55 ] is the measured energy spectrum due to external @xmath8fe calibration sources ( 5.90 and 6.49 kev ) together with x - rays from ti ( 4.51 and 4.93 kev ) , ca ( 4.01 kev ) , s ( 2.46 kev ) and al ( 1.55 kev ) .
random trigger events uncorrelated to the detector provided the zero - energy pedestals .
pulse shape discrimination ( psd ) criteria were applied as illustrated in figure [ psd ] by correlating two output with different electronics amplifications and shaping times .
the electronic noise edge was suppressed by psd and a threshold of 100 ev was achieved .
the deviations of the low energy spectra of the selected events from a flat distribution gave the efficiencies of the psd cuts .
pulse shape discrimination : correlations of signals with different electronic amplifications and shaping times lead to suppression of the noise edge . ]
pulse shape discrimination : correlations of signals with different electronic amplifications and shaping times lead to suppression of the noise edge . ]
low - background data were taken at ks under the same shielding configurations as the magnetic moment experiments @xcite , using a 4-channel ulege detector each having an active mass of 5 g. the recorded spectrum with 0.34 kg - day of data after cosmic - rays and anti - compton vetos and psd selection is displayed in figure [ bkgspect ] .
it can be seen that comparable background level was achieved as the cresst - i @xcite experiment with 1.51 kg - day of data .
a summary of the spin - independent exclusion plot is depicted in figure [ dmplot ] .
comparable limits to cresst - i can be expected from the current ks results .
intensive efforts on the data analysis are underway . the projected sensitivities with 1 kg - year of data at 100 ev threshold and a 1 cpd background level are shown . expected spin - independent sensitivity region for cdm searches using a ulege detector with a mass of 1 kg at 1 cpd background level . also shown are the present exclusion boundaries @xcite and the dama allowed regions @xcite .
the striped region is the parameter space favored by the susy models @xcite . ]
expected spin - independent sensitivity region for cdm searches using a ulege detector with a mass of 1 kg at 1 cpd background level .
also shown are the present exclusion boundaries @xcite and the dama allowed regions @xcite .
the striped region is the parameter space favored by the susy models @xcite . ]
an r&d program towards the realizations of a full - scale experiment is rigorously pursued .
quenching factor measurement for nuclear recoils in ge with sub - kev ionization energy will be performed at a neutron beam facility .
background studies are conducted at both ks and the yang - yang underground laboratory ( 700 mwe ) in south korea with the various prototypes .
background understanding at the sub - kev range is a challenging and unexplored subject in its own right .
external background is expected to be reduced due to self - shielding effects in a kg - mass detector , as well as by additional active veto ge layers enclosing hermetically the ulege inner target @xcite .
studies are under way with a 180-g segmented ulege prototype equipped with a veto ring and dual - readout channels from both the signal and high - voltage electrodes . a 500-g detector similar to the design demonstrated in ref .
@xcite is being constructed . |
magnetic lattices consisting of periodic arrays of current - carrying wires @xcite or permanent magnetic films @xcite have recently been proposed as an alternative approach to optical lattices @xcite for trapping and manipulating small clouds of ultracold atoms and quantum degenerate gases , including bose - einstein condensates .
magnetic lattices may be considered as complementary to optical lattices , in much the same way as magnetic traps are complementary to optical dipole traps : they do not require ( intense and stable ) laser beams and there is no decoherence or light scattering due to spontaneous emission ; they can produce highly stable and reproducible potential wells leading to high trap frequencies ; and only atoms in low magnetic field seeking states are trapped , thus allowing the possibility of performing rf evaporative cooling @xmath0 @xmath1 in the lattice and the study of very low temperature phenomena in a periodic lattice .
simple 1d magnetic lattices consisting of periodic arrays of traps or waveguides have been constructed using both current carrying wires @xcite and permanent magnets @xcite on atom chips .
+ in a recent paper , we proposed a 2d magnetic lattice consisting of two crossed layers of periodic arrays of parallel rectangular magnets @xcite . in this paper
, we report on a new class of configurations of permanent magnets comprising @xmath2 periodic arrays of square magnets of different thickness for producing 1d and 2d arrays of magnetic microtraps with non - zero potential minima and variable barrier height .
analytical expressions are presented for various physical quantities and compared with numerical calculations . in one of the configurations of magnets , we show that by varying the angle between the crossed arrays of magnets it is possible to go from a 1d array of microtraps to a 2d array of microtraps .
the 2d magnetic lattices consisting of four periodic arrays of square magnets of different thickness reported here may prove easier to implement experimentally than the crossed periodic arrays of parallel rectangular magnets proposed previously @xcite .
( a)-(b ) shows a configuration of four periodic arrays of square magnetic slabs with thicknesses @xmath3 , @xmath4 , @xmath5 and @xmath6 , respectively . in , we find the components of the magnetic field due to a single array of square magnetic slabs and then use our results to obtain the total magnetic field due to the four arrays of square magnetic slabs and an external bias field @xmath7 .
@xmath8{figure1a.eps } & \includegraphics[angle=0,width=6cm]{figure1b.eps}\\ \includegraphics[angle=0,width=6cm]{figure1c.eps } & \includegraphics[angle=0,width=6cm]{figure1d.eps } \end{array}$ ] here , we consider a single infinite periodic array of square magnetic slabs of thickness @xmath9 , at a distance @xmath10 from the plane @xmath11 , with periodicity @xmath12 along the @xmath13- and @xmath14-directions and perpendicular magnetization @xmath15 [ ( c ) ] .
the sum of this configuration and a similar one displaced by @xmath16 along the @xmath13-axis gives an infinite periodic array of parallel rectangular magnets with periodicity @xmath12 along the @xmath14-direction and the same magnetization @xmath15 and thickness @xmath9 [ ( d ) ] . for distances from the surface which are large compared with @xmath17 ,
the components of the magnetic field are @xcite @xmath18 @xmath19 @xmath20 where @xmath21 , @xmath22 and @xmath23 ( gaussian units ) . considering the corresponding surface currents around the magnetic slabs in ( d ) , we find that the array of blue ( grey in grey style ) square magnets produces the same component of magnetic field along the @xmath14-axis as the array of red ( black in grey style ) magnets .
thus , the @xmath14-component for the array of blue ( or red ) magnets can be written as half that of .
furthermore , ( c ) is symmetrical with respect to @xmath13 and @xmath14 , and therefore , for this infinite array of square magnets , the dependence of the @xmath13-component of the magnetic field on @xmath13 is the same as the dependence of the @xmath14-component of the magnetic field on @xmath14 . on the other hand
, the @xmath24-component of the magnetic field should be symmetrical with respect to @xmath13 and @xmath14 and also needs to be maximum at @xmath25 .
thus the components of the magnetic field for the configuration of magnets in ( c ) may be written as @xmath26 @xmath27 @xmath28 { \rm e}^{-kz}\ ] ] where @xmath29 and @xmath30 . using - , we can write the components of the magnetic field due to the four arrays of square permanent magnetic slabs [ ( a)-(b ) ] with external bias field @xmath31 as @xmath32 @xmath33 @xmath34 { \rm e}^{-kz}\ ] ] where @xmath35 and @xmath36 @xmath37 @xmath38 the magnitude of the magnetic field above the magnetic arrays is then @xmath39{\rm e}^{-kz } \nonumber \\ & + & [ b_{0x}^2+b_{0y}^2 + 2 b_{0x}b_{0y } \cos(kx)\cos(ky ) ] { \rm e}^{-2kz } \bigr\}^{1\over 2 } \label{e5}\end{aligned}\ ] ] this configuration of square magnets gives a 2d periodic lattice of magnetic traps with _ non - zero _ potential minima given by @xmath40 which are located at @xmath41 @xmath42 @xmath43 if we require a 2d lattice with _ non - zero _ magnetic field minima , and impose the constraints @xmath44 @xmath45 on the geometrical parameters and the components of the bias magnetic field , where according to - , @xmath46 and @xmath47 are independent of the magnetization and depend only on the geometrical constants @xmath48 and @xmath49 .
+ the potential barrier heights in the three directions are given by @xmath50^{1\over 2 } \hspace{-.2cm}- b_{min } , \hspace{.1 cm } j=1,2\ ] ] @xmath51 where @xmath52 and @xmath53 .
furthermore , the curvatures of the magnetic field at the centre of the traps can be written as @xmath54 @xmath55 the trap frequencies , for an atom in hyperfine state @xmath56 with magnetic quantum number @xmath57 , are given by @xmath58^{1\over 2 } , \quad j=1,2\ ] ] @xmath59 where @xmath60 = @xmath61 , @xmath62 is the land g - factor , @xmath63 is the bohr magneton and @xmath64 is the atomic mass . to create a 2d lattice that is symmetrical with respect to @xmath13 and @xmath14
we impose the constraint @xmath65 from which we obtain @xmath66 , where @xmath67 , @xmath68 and @xmath69 .
this is the same as the condition given in @xcite for two crossed layers of infinite periodic arrays of magnets with bias fields ( ( a ) of @xcite ) ; so our results are the same as the equations given in @xcite . here again , our analytical expressions for @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 and @xmath80 depend on the @xmath13-component of the bias field @xmath81 only , rather than on both @xmath81 and @xmath82 : @xmath83 @xmath84 @xmath85 @xmath86 @xmath87 where @xmath46 and @xmath47 are given by and and @xmath88 , @xmath89 , @xmath90 are dimensionless constants which depend on @xmath91 and are given by @xmath92 @xmath93 @xmath94 @xmath95 @xmath96 for a symmetrical 2d magnetic lattice with non - zero potential minima above the surface of the top array , we have the constraints which are the same as in @xcite @xmath97 in this section , we consider special cases of the magnetic lattices introduced in . when @xmath98 , @xmath99 and @xmath100 [ ( d ) ] , we obtain an infinite periodic array of rectangular magnets and - give @xmath101 @xmath8{figure2a.eps } & \includegraphics[angle=0,width=6cm]{figure2b.eps } \end{array}$ ] substituting @xmath46 and @xmath47 from into - ,
we obtain expressions for @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath102 , @xmath103 , @xmath78 , @xmath79 and @xmath80 which are the same as in @xcite . here
, we consider a configuration of square magnets in which @xmath104 , @xmath105 , @xmath106 and @xmath107 [ ( a ) ] . from -
, we have @xmath108 is in agreement with the expressions given in @xcite where this magnetic lattice has been studied in detail .
@llll parameter & definition & numerical inputs & analytical inputs + ` n_{sq } ` & number of squares & @xmath109 & @xmath110 + ` in $ x$- or $ y$-direction ` + ` a\hspace{.1cm}(\mu m ) ` & period of magnetic lattice & @xmath111 & @xmath111 + ` l_x = l_y\hspace{.1cm}(\mu m ) ` & length of the chip & @xmath112 & @xmath110 + ` along $ x$ or $ y$ ` & + ` t_1\hspace{.1cm}(\mu m ) ` & thickness of magnetic & @xmath113 & @xmath113 + ` film ( first ) ` & + ` t_2\hspace{.1cm}(\mu m ) ` & thickness of magnetic & @xmath114 & @xmath114 + ` film ( second ) ` & + ` t_3\hspace{.1cm}(\mu
m ) ` & thickness of magnetic & @xmath115 & @xmath115 + ` film ( third ) ` & + ` 4 \pi m_z\hspace{.1cm}(g ) ` & magnetization along z & @xmath116 & @xmath116 + ` b_{1x}\hspace{.1cm}(g ) ` & bias magnetic field along @xmath13 & @xmath117 @xmath118 & @xmath117 @xmath118 + ` b_{1y}\hspace{.1cm}(g ) ` & bias magnetic field along @xmath14 & @xmath117 @xmath119 & @xmath117 @xmath119 + ` b_{1z}\hspace{.1cm}(g ) ` & bias magnetic field along @xmath24 & @xmath117 @xmath120 & @xmath121 + @llll parameter & definition & numerical results & analytical results + ` x_{min}\hspace{.1cm}(\mu m ) ` & @xmath13 co - ordinate of potential minimum & @xmath122 & @xmath122 + ` y_{min}\hspace{.1cm}(\mu m ) ` & @xmath14 co - ordinate of potential minimum & @xmath122 & @xmath123 + ` z_{min}\hspace{.1cm}(\mu m ) ` & @xmath24 co - ordinate of potential minimum & @xmath124 & @xmath124 + ` d\hspace{.1cm}(\mu m ) ` & distance of potential & @xmath125 & @xmath125 + ` minimum from surface ` @xmath126 + ` b_{min}\hspace{.1cm}(g ) ` & magnetic field at potential minimum & @xmath127 & @xmath128 + ` \frac{\partial^2 b}{\partial x^2}\hspace{.1cm}({g\over{cm}^2 } ) ` & curvature of @xmath129 along @xmath13 & @xmath130 & @xmath130 + ` \frac{\partial^2 b}{\partial y^2}\hspace{.1cm}({g\over{cm}^2 } ) ` & curvature of @xmath129 along @xmath14 & @xmath130 & @xmath130 + ` \frac{\partial^2 b}{\partial z^2}\hspace{.1cm}({g\over{cm}^2 } ) ` & curvature of @xmath129 along @xmath24 & @xmath131 & @xmath131 + ` \delta b^x\hspace{.1cm}(g ) ` & magnetic barrier height along @xmath13 & @xmath132 & @xmath132 + ` \delta b^y\hspace{.1cm}(g ) ` & magnetic barrier height along @xmath14 & @xmath133 & @xmath132 + ` \delta b^z\hspace{.1cm}(g ) ` & magnetic barrier height along @xmath24 & @xmath134 & @xmath134 + ` u_{min}/k_b \hspace{0.1cm}(\mu { k } ) ` & potential energy minimum & @xmath135 & @xmath136 + ` \delta u^x / k_b \hspace{0.1cm}(\mu { k } ) ` & potential barrier height along @xmath13 & @xmath137 & @xmath137 + ` \delta u^y / k_b \hspace{0.1cm}(\mu { k } ) ` & potential barrier height along @xmath14 & @xmath137 & @xmath137 + ` \delta u^z / k_b \hspace{0.1cm}(\mu { k } ) ` & potential barrier height along @xmath24 & @xmath138 & @xmath138 + ` \omega_x/ 2\pi \hspace{0.1 cm } ( { kh}z ) ` & trap frequency along @xmath13 & @xmath139 & @xmath139 + ` \omega_y/ 2\pi \hspace{0.1 cm } ( { kh}z ) ` & trap frequency along @xmath14 & @xmath139 & @xmath139 + ` \omega_z/ 2\pi \hspace{0.1 cm } ( { kh}z ) ` & trap frequency along @xmath24 & @xmath140 & @xmath141 + ` \hbar\omega_x / k_b \hspace{0.1cm}(\mu { k } ) ` & level spacing along @xmath13 & @xmath142 & @xmath142 + ` \hbar\omega_y / k_b \hspace{0.1cm}(\mu { k } ) ` & level spacing along @xmath14 & @xmath142 & @xmath142 + ` \hbar\omega_z / k_b \hspace{0.1cm}(\mu { k } ) ` & level spacing along @xmath24 & @xmath143 & @xmath143 + when @xmath144 [ ( b ) ] , from - , we have @xmath145 we note that , according to , for a symmetrical 2d lattice with non - zero magnetic potential minima , we should have @xmath146 .
using equations - and we can determine analytically the relevant quantities for this system .
this magnetic lattice has already been studied numerically @xcite .
gives the numerical @xcite and analytical inputs and gives calculated values for the relevant quantities .
there is good agreement between the analytical results for an infinite lattice and the numerical results @xcite which were obtained in the central region of the lattice . in ( b ) , if we take @xmath147 and @xmath148 , we have a chessboard configuration of square magnetic slabs . in this case , from , we obtain @xmath149 considering , we do not have a lattice of _ non - zero _ magnetic field minima .
in fact , we do not have a magnetic lattice , because according to - we obtain a uniform magnetic field if we look at points sufficiently far from the surface of the array .
@xmath150{figure3.eps } \end{array}$ ]
here , we consider a configuration of magnets which is a generalization of
. shows this configuration of rectangular magnetic slabs in which the upper array of parallel magnets is rotated at an arbitrary angle @xmath151 about the @xmath24 axis which is normal to the surface .
the magnets in the lower array are parallel to the @xmath13-axis while the magnets in the upper array are parallel to the @xmath152-axis .
if @xmath153 , according to , we have the two crossed infinite periodic arrays of magnets shown in ( a ) .
considering the symmetry , in the rotated coordinate system of reference , we have for the upper array @xmath154 @xmath155 @xmath156 where @xmath157 , @xmath158 and @xmath159 .
now , we can write the components of the magnetic field in the original coordinate system @xmath160 @xmath161 @xmath162 the components of the total magnetic field due to both arrays of parallel magnets plus a bias field @xmath163 are then @xmath164 @xmath165 { \rm e}^{-k{{z}}}+b_{{1y}}\ ] ] @xmath166{\rm e}^{-k{{z}}}\ ] ] where @xmath167 and @xmath168 .
the magnitude of the magnetic field is then @xmath169{\rm e}^{-kz}\nonumber\\ & + & ( b_{0y^{\prime}}^2+b_{0y}^2 + 2 b_{0y^{\prime}}b_{0y } [ \cos(ky^{\prime})\cos(ky)\nonumber\\ & + & \sin(ky^{\prime})\sin(ky)\cos\theta]){\rm e}^{-2kz } \bigr\}^{1\over 2}\label{e22}\end{aligned}\ ] ] if @xmath170 , these arrays of rectangular magnets can give a 2d periodic lattice of magnetic traps with _ non - zero _ potential minima given by @xmath171 the potential minima are located at @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 the period @xmath177 of the magnetic potential minima , which are located along straight lines parallel to the @xmath13- and @xmath152-axes ( defined in ) , varies with @xmath151 and is given by @xmath178 if @xmath151 changes between @xmath121 and @xmath179 , the period of the magnetic lattice varies between @xmath180 and the minimum period @xmath12 , which is the period of the magnetic slabs along the @xmath14- and @xmath181-axes .
@xmath182{figure4.eps } \end{array}$ ] .
now , we can write the constraints for a lattice with _ non - zero _ magnetic field minima @xmath183 @xmath184 according to - and the contour plots of the magnetic field for different values of @xmath151 ( ) , apart from @xmath185 for which we have a 1d array of 2d magnetic microtraps , we have a 2d array of 3d microtraps with different periods in the @xmath13- and @xmath152-directions , as @xmath151 varies .
looking at more carefully , we find that as @xmath151 changes the distance between the magnetic traps along the interior bisector of the angle @xmath151 between the positive @xmath14- and @xmath181-axes also changes .
thus , depending on the barrier heights , the tunnelling rate between the traps in this direction will be higher than the tunnelling rate in the @xmath13- and @xmath152-directions . here
, we give the barrier heights in the @xmath13- , @xmath152- and @xmath24-directions @xmath186 ^ 2 \nonumber\ ] ] @xmath187 ^ 2 \bigr\}^{1\over 2}\ ] ] @xmath188 @xmath189 where , @xmath52 , @xmath190 and by definition we have @xmath191 @xmath192 @xmath193 where @xmath194 @xmath195 we also have @xmath196 @xmath197 furthermore , the curvatures are given by @xmath198 @xmath199 @xmath200 @xmath201\ ] ] @xmath202 @xmath203 @xmath204\ ] ] @xmath205 @xmath206 @xmath207 @xmath208\ ] ] @xmath209 @xmath210 @xmath211 @xmath212\ ] ] @xmath213 where @xmath214 @xmath215 @llll parameter & definition & numerical inputs & analytical inputs + ` n_{r } ` & number of parallel rectangular & @xmath216 & @xmath110 + ` magnets in $ x$- or $ x^{\prime}$-direction ` + ` a\hspace{.1cm}(\mu m ) ` & period of magnetic lattice & @xmath111 & @xmath111 + ` l_x = l_x^{\prime}\hspace{.1cm}(\mu m ) ` & length of the magnets & @xmath217 & @xmath110 + ` along $ x$- or $ x^{\prime}$-direction ` & + ` t_1\hspace{.1cm}(\mu m ) ` & thickness of magnetic & @xmath218 & @xmath218 + ` slabs parallel to the $ x$-direction ` & + ` t_2\hspace{.1cm}(\mu m ) ` & thickness of magnetic & @xmath219 & @xmath219 + ` slabs parallel to the $ x^{\prime}$-direction ` & + ` s_1\hspace{.1cm}(\mu m ) ` & distance from the plane & @xmath220 & @xmath220 + ` $ z=0 $ to the lower surface of ` & + ` the magnets parallel ` & + ` to the $ x$-direction ` & + ` s_2\hspace{.1cm}(\mu m ) ` & distance from the plane & @xmath221 & @xmath221 + ` $ z=0 $ to the lower surface of ` & + ` the magnets parallel ` & + ` to the $ x^{\prime}$-direction ` & + ` 4 \pi m_z\hspace{.1cm}(g ) ` & magnetization along z & @xmath116 & @xmath116 + ` b_{1x}\hspace{.1cm}(g ) ` & bias magnetic field along @xmath13 & @xmath117 @xmath222 & @xmath117 @xmath223 + ` b_{1y}\hspace{.1cm}(g ) ` & bias magnetic field along @xmath14 & @xmath117 @xmath224 & @xmath117 @xmath224 + ` b_{1z}\hspace{.1cm}(g ) ` & bias magnetic field along @xmath24 & @xmath117 @xmath225 & @xmath226 + @llll parameter & definition & numerical results & analytical results + @xmath227 & @xmath13 co - ordinate of potential minimum & @xmath228 & @xmath228 + @xmath229 & @xmath14 co - ordinate of potential minimum & @xmath122 & @xmath123 + @xmath230 & @xmath24 co - ordinate of potential minimum & @xmath231 & @xmath231 + @xmath232 & distance of potential & @xmath233 & @xmath233 + & minimum from surface & @xmath126 + @xmath234 & magnetic field at potential minimum & @xmath235 & @xmath236 + @xmath237 & curvature of @xmath129 along @xmath13 & @xmath238 & @xmath238 + @xmath239 & curvature of @xmath129 along @xmath14 & @xmath240 & @xmath241 + @xmath242 & curvature of @xmath129 along @xmath24 & @xmath243 & @xmath243 + @xmath244 & magnetic barrier height along @xmath13 & @xmath245 & @xmath246 + @xmath247 & magnetic barrier height along @xmath152 & @xmath248 & @xmath246 + @xmath249 & magnetic barrier height along @xmath24 & @xmath250 & @xmath250 + @xmath251 & potential energy minimum & @xmath252 & @xmath253 + @xmath254 & potential barrier height along @xmath13 & @xmath255 & @xmath255 + @xmath256 & potential barrier height along @xmath152 & @xmath257 & @xmath255 + @xmath258 & potential barrier height along @xmath24 & @xmath259 & @xmath260 + @xmath261 & trap frequency along @xmath13 & @xmath262 & @xmath262 + @xmath263 & trap frequency along @xmath14 & @xmath264 & @xmath264 + @xmath265 & trap frequency along @xmath24 & @xmath266 & @xmath267 + @xmath268 & level spacing along @xmath13 & @xmath269 & @xmath269 + @xmath270 & level spacing along @xmath14 & @xmath271 & @xmath271 + @xmath272 & level spacing along @xmath24 & @xmath273 & @xmath273 + @xmath274{figure5.eps } \end{array}$ ] for @xmath275 , if we have @xmath276 , where @xmath277 , then the barrier heights in the @xmath13- and @xmath152-directions are equal .
thus , we have a symmetrical 2d lattice of 3d microtraps in the @xmath13- and @xmath152-directions and , as shows , the microtraps in the @xmath13- and @xmath152-directions have the same period .
this figure shows our analytical results for an infinite lattice , which are in excellent agreement with our numerical calculations in the central region of the lattice .
for the numerical calculations we have written a mathematica code and have used the software package radia @xcite interfaced to mathematica . here
, the quantities of interest are @xmath278 @xmath279 @xmath280 @xmath281 @xmath282 @xmath283 @xmath284 @xmath285 @xmath286 where @xmath287 and @xmath288 , @xmath89 , @xmath289 are dimensionless parameters which depend on @xmath290 and @xmath151 . using the definitions @xmath291 , @xmath292 , @xmath293 and @xmath294
, we have @xmath295 @xmath296^{1\over 2}}}}\ ] ] @xmath297 @xmath298^{3\over 2}}\ ] ] @xmath299\sin^2\theta}{|c_1h_1(\theta)|[h_2(\theta)]^{3\over 2}}\ ] ] @xmath300^{1\over 2}}\ ] ] @xmath301^{1\over 2}}{|h_1(\theta)h_2(\theta)|}-\beta_1(\theta)\ ] ] @xmath302^{1\over 2}-\beta_1(\theta)\ ] ] subject to @xmath276 , the conditions for _ non - zero _ magnetic potential minima are @xmath303 @xmath304 according to - , and , for a constant @xmath305 , it is possible to change the trap frequencies by varying @xmath151 and/or the bias magnetic field @xmath306 .
furthermore , as , , and indicate , we can adjust the magnetic field potential barriers by changing the angle @xmath151 , the bias magnetic field @xmath306 , or both .
+ + gives the numerical and analytical inputs and gives calculated values for the relevant quantities .
there is good agreement between the analytical results for an infinite lattice and the numerical results which were obtained in the central region of the lattice .
@xmath307{figure6.eps } \end{array}$ ]
here , we consider a special configuration involving three periodic arrays of parallel rectangular magnets . the lower array ( array 1 ) and the middle array ( array 2 ) are at an angle @xmath151 with respect to each other , similar to the crossed arrays shown in , while the rectangular magnets in the top array ( array 3 ) are parallel to those of the middle array and remain parallel as @xmath151 changes .
the period @xmath308 is twice the periods @xmath309 and @xmath310 . to obtain the components of the magnetic field due to these three arrays of magnets plus bias field @xmath31 , we add the magnetic field due to array 3 to that of arrays 1 and 2 [ - ] : @xmath311 @xmath312 { \rm e}^{-k{{z}}}+ [ b_{0{{y}^{\prime\prime}}}\cos\theta \sin ( k{{y}^{\prime}}/2 ) ] { \rm e}^{-k{{z/2}}}+ b_{{1y}}\ ] ] @xmath313{\rm e}^{-k{{z } } } + b_{0{{y}^{\prime\prime } } } \cos ( k{{y}^{\prime}}/2){\rm e}^{-k{{z/2}}}\ ] ] where @xmath314 and @xmath315 , @xmath316 and @xmath317 . this magnetic field
is shown for @xmath318 and @xmath319 in ( a)-(c ) and ( d)-(f ) , respectively .
the period @xmath320 @xmath321 , which is twice the periods @xmath309 and @xmath310 .
the thicknesses of the layers are @xmath322 @xmath323 , @xmath324 @xmath323 and @xmath325 @xmath323 and the lower surfaces of the layers are at distances @xmath326 , @xmath327 @xmath323 and @xmath328 @xmath323 from the plane @xmath11 .
the magnetization is as before and the components of the bias magnetic field are @xmath329 @xmath330 , @xmath331 @xmath330 and @xmath332 . according to
, we can change the separation and also the barrier height between adjacent microtraps by varying @xmath151 .
this might be of interest for a two - qubit quantum gate process @xcite .
we have introduced a class of permanent magnetic lattices for ultracold atoms and quantum degenerate gases .
the potential barriers between the adjacent microtraps can be altered by varying the bias magnetic field , as indicated by the relevant equations .
thus , these configurations of permanent magnets together with an adjustable bias magnetic field could be used to study quantum tunnelling and to realize a bec to mott insulator quantum phase transition for quantum information processing in a magnetic lattice .
+ we have shown that for a configuration of two crossed periodic arrays of parallel rectangular magnets with a variable angle between the two arrays , the magnetic lattice goes from a 1d to a 2d lattice as the angle @xmath151 between the arrays is changed from @xmath121 to @xmath333 .
this may prove useful for loading ultracold atoms into a 2d lattice of 3d microtraps by initially loading the atoms into the 1d lattice of 2d microtraps and then by changing the angle from @xmath334 to @xmath335 . also , by varying the angle @xmath151 , it is possible to change the depth of the traps adiabatically , which suggests the possibility of realizing a ` mechanical ' bec to mott insulator quantum phase transition in a magnetic lattice .
moreover , the period of the magnetic lattice can be adjusted as the angle @xmath151 is changed .
this could have applications in quantum information processing and also in creating new artificial crystalline materials . in a more general case
, it is possible to have two different periods @xmath336 and @xmath337 in the @xmath14- and @xmath181-directions , respectively , to allow for a more flexible magnetic lattice .
a configuration of magnets with three periodic arrays of rectangular magnets having two different periods @xmath338 and @xmath339 may be useful for implementing a two - qubit quantum gate @xcite in a magnetic lattice .
10 yin j , gao w , hu j and wang y 2002 _ opt .
commun . _ * 206 * 99 grabowski a and pfau t 2003 _ eur .
j. d _ * 22 * 347 hinds e a and hughes i g 1999 r119 ghanbari s , kieu t d , sidorov a and hannaford p 2006 847 jaksch d , bruder c , cirac j i , gardiner c w and zoller p 1998 3108 greiner m , mandel o , esslinger t , hnsch t and bloch i 2002 _ nature _ * 415 * 39 gnther a , kraft s , kemmler m , koelle d , kleiner r , zimmermann c and fortgh j 2005 170405 hnsel w , hommelhoff p , hnsch t w and reichel j 2001 _ nature _ * 413 * 498 barb i , gerritsma r , xing y t , goedkoop j b and spreeuw r j c 2005 _ eur .
j d _ * 35 * 75 sinclair c d j , curtis e a , llorente - garcia i , retter j a , hall b v , eriksson s , sauer b e and hinds e a 2005 _ phys rev . a _ * 72 * 031603 sidorov a i , mclean r j , scharnberg f , gough d s , davis t j , sexton b a , opat g i and hannaford p 2002 _ acta phys .
polonica b _ * 33 * 2137 boyd m , streed e w , medley p , campbell g k , mun j , ketterle w and pritchard d e arxiv : cond - mat/0608370 v1 16 aug 2006 available from : http://www.esrf.fr/machine/support/ids/public/index.html barenco a , bennett c h , cleve r , divicenzo d p , margolus n , shor p , sleator t , smolin j a and weinfurter h 1995 _ phys rev .
a _ * 52 * 3457 | we report on a class of configurations of permanent magnets on an atom chip for producing 1d and 2d periodic arrays of magnetic microtraps with non - zero potential minima and variable barrier height for trapping and manipulating ultracold atoms and quantum degenerate gases .
we present analytical expressions for the relevant physical quantities and compare them with our numerical results and with some previous numerical calculations . in one of the configurations of permanent magnets ,
we show how it is possible by changing the angle between the crossed periodic arrays of magnets to go from a 1d array of 2d microtraps to a 2d array of 3d microtraps and thus to continuously vary the barrier heights between the microtraps .
this suggests the possibility of performing a type of ` mechanical ' bec to mott insulator quantum phase transition in a magnetic lattice .
we also discuss a configuration of magnets which could realize a two - qubit quantum gate in a magnetic lattice . |
i thank s. godfrey and w. venus for discussions and helpful comments .
r. barate et al . , phys .
* 294 * ( 1998 ) 1 .
( 1999 ) 364 .
( 1996 ) 61 .
m. acciarri et al . , ( 1997 ) 465 .
r. akers et al . , ( 1995 )
k. ackerstaff et al . , ( 1998 ) 19 .
k. ackerstaff et al . , ( 1998 ) 411 .
particle data group , ( 2000 ) 1 .
t. sjstrand , comp .
phys . comm . * 82 * ( 1994 ) 74 .
yi - jin pei , ( 1996 ) 39 .
chliapnikov , ( 1999 ) 263 .
chliapnikov , ( 2001 ) 18 .
k. ackerstaff et al . , ( 1997 ) 425 .
r. barate et al . , ( 2000 ) 597 . ( 2000 ) 209 .
k. ackerstaff et al . , ( 1998 ) 439 . n. isgur and and m.b .
wise , ( 1989 ) 113 ; * b237 * ( 1990 ) 527 .
d. buskulic et al . , ( 1996 ) 393 .
( 1995 ) 598 .
m. acciarri et al . , ( 1999 ) 323 .
k. akers et al . , ( 1995 )
r. barate et al . , ( 1998 ) 215 .
cdf collab .
, t. affolder et al . , ( 2001 ) 072002 .
g. abbiendi et al . , cern - ep/2000 - 125 .
s. godfrey and n. isgur , ( 1985 ) 189 .
eichten , c.t . hill and c. quigg , ( 1993 ) 4116 .
m. grounau , and j.l .
rosner , ( 1994 ) 254 .
s.n . gupta and j.m .
jonson , ( 1995 ) 168 . n. isgur , ( 1998 ) 4041 .
d. ebert , v.o .
galkin and r.n .
faustov , ( 1998 ) 5663 .
( 1995 ) 353 .
m. acciarri et al . , ( 1995 ) 589 .
k. ackerstaff et al . , ( 1997 ) 413 .
uvarov , ( 2000 ) 10 ; * b511 * ( 2001 ) 136 .
accmor collab .
, c. daum et al . , nucl .
b187 * ( 1981 ) 1 .
gubankova , phys . of atomic nuclei ,
* 58 * ( 1995 ) 660 .
blundell , s. godfrey and b. phelps , ( 1996 ) 3712 .
chliapnikov , ( 2000 ) 129 .
anisovich and a.v .
sarantsev , ( 1997 ) 137 . | the production rates of primary vector and @xmath0-wave mesons in z@xmath1 hadronic decays are analysed .
the mass dependence of production rates for the bottom , charm , strange charm and three families of the light - flavour mesons is found to be very similar , allowing to relate the relative production rates for mesons with different flavours and , possibly , their masses .
the strange axial mesons k@xmath2 and k@xmath3 might be assigned to the @xmath4 and @xmath5 levels degenerate with the @xmath6 and @xmath7 levels of the k@xmath8 and k@xmath9 , respectively , if the observed k@xmath8 mass is replaced by its `` bare '' @xmath10 mass corresponding to the @xmath11-matrix pole and close to the @xmath12 mass .
then the @xmath6 and @xmath4 levels are _ below _ the @xmath5 and @xmath7 levels for the strange , charm and bottom mesons .
the lep experiments accumulated rich information on inclusive production of the light - flavour , charm and bottom mesons in the @xmath13 hadronic decays including data on @xmath0-wave meson production . in this letter , we use these data to compare the production of primary vector and @xmath0-wave mesons in an attempt to relate the production of these states for light and heavy flavours . the total production rates of the vector ^0 @xmath14 , @xmath15 , ^*0(892 ) @xmath16 and @xmath17 , the tensor f_2(1275 ) @xmath18 , _
2^*0(1430)@xmath19 and f_2(1525 ) @xmath20 , and the scalar f_0(980 ) @xmath21 and a^+_0(980 ) @xmath22 mesons measured by the lep experiments @xcite are presented in table 1 . for the vector and scalar mesons ,
the measurements from the different lep experiments agree within errors .
therefore subsequently we used the rates obtained by averaging the results of these experiments , also presented in table 1 .
.the total production rates of the vector , tensor and scalar mesons in the light - quark sector measured by the lep experiments , averaged total rates for the vector and scalar mesons , fractions of primary mesons obtained from the jetset model and direct rates determined by multiplying the total rates by the fractions of primary mesons . for the ^*(892 ) @xmath23 , _
2^*(1430 ) @xmath24 and a^+_0(980 ) @xmath22 antiparticles and charge conjugates are not included into the definition of the rates . [ cols="<,^,^,^,^",options="header " , ] the relative ^ * @xmath25 production rate in @xmath26-quark jet , @xmath27 , was measured by the lep experiments for a mixture of the states @xmath28 , @xmath29 and @xmath30 with the following results : @xmath31 @xcite , @xmath32 @xcite , @xmath33 @xcite and @xmath34 @xcite , with the averaged value of @xmath35 . assuming that @xmath36 are produced for each @xmath37
, we obtain _ ^ * @xmath25(u , d)/_b- = 0.580 0.032 . for determining the rate of the promptly produced @xmath38 , the decays of @xmath0-wave mesons into ^ * @xmath25@xmath39 have to be taken into account . for this , the branching fractions into ^ * @xmath25@xmath39 shown in table 2 were used ( the same as in @xcite ) . with the relative production rates from table 3 calculated using modified total spin counting , this yields @xmath40 and @xmath41 for the l3 and opal results respectively .
the additional systematic errors account to half of the difference between these values and the value @xmath42 found by opal @xcite .
the resulting relative rates of the promptly produced @xmath38 are presented in table 3 , together with similarly obtained values based on the l3 results with simple total spin counting and opal results with the opal proportion of the relative rates . as one
can see from table 3 , the relative rates of the four spin states , and especially the _ 0 @xmath43/_2^ * @xmath44 ratio , are quite sensitive to the spin counting rules assumed . for the masses of these states from opal
, this ratio obtained using the opal proportion of the rates is larger by a factor of 5.2 than the same ratio obtained with modified total spin counting .
this suggests that significantly different fitted values of the masses might be obtained if modified total spin counting were applied instead of the opal proportion . on the other hand ,
the rate of promptly produced ^ * @xmath25 is practically insensitive to the difference in the counting rules .
the _ 2^ * @xmath44 rates obtained with modified spin counting at the masses from the l3 and opal are consistent within 1.7 standard deviations ( or even less since the _
2^ * @xmath44 mass from l3 is larger than from opal ) .
this suggests that the production rates of the _
2^ * @xmath44 and promptly produced ^ * @xmath25 are sufficiently reliable to allow comparison of their mass dependence with other data .
the relative rates of ^*_1 @xmath45 , _
2^ * @xmath44 and promptly produced ^ * @xmath25 in @xmath26-quark jets calculated using modified total spin counting and divided by the spin counting factors @xmath46 are shown at the ^*_1 @xmath45 and _ 2^ * @xmath44 masses from l3 in fig .
1a and at the ^*_1 @xmath45 and _ 2^ * @xmath44 masses from opal in fig .
1b .
the fits of the data to six exponentials with different normalization parameters for the six meson families , but the _ same _ slope parameter @xmath47 @xmath48 in fig .
1a and @xmath49 @xmath48 in fig .
1b describe the data well ( solid lines in fig . 1 ) .
the slope parameters are very close to the value @xmath50 @xmath48 obtained for the light - flavour mesons .
thus we see that the mass dependences of the production rates per spin projection are indeed very similar for the light - flavour , charm and bottom mesons and _ 2^ * @xmath44 , since for the ^*_1 @xmath45 and _ 2^ * @xmath44 rates the same mass dependence as for the light - flavour mesons has been imposed by eqs .
( 8) and ( 9 ) . ] .
one important lesson from this observation is the existence of a close relationship between the masses of the @xmath0-wave states and their production rates . if the masses of the @xmath51 states with @xmath52 and @xmath53 are below ( above ) the masses of the @xmath54 states with @xmath55 and @xmath56 , their production rates per spin projection are larger ( smaller ) than for the @xmath57 states , as shown for the ^*_1 @xmath45 and _ 2^ * @xmath44 in fig .
1a ( fig .
1b ) .
apart from the @xmath0-wave mesons discussed above , only the production rates of the scalars a^+_0(980 ) @xmath22 and f_0(980 ) @xmath21 were measured at lep @xcite .
they are presented in table 1 and fig . 1 . the a^+_0(980 )
@xmath22 rate is consistent within errors with the mass dependence of the ^0 @xmath14 , @xmath15 and f_2(1275 ) @xmath18 rates .
the f_0(980 ) @xmath21 rate is consistent with the a^+_0(980 ) @xmath22 rate as expected , but appears to be slightly higher than follows from the mass dependence of the ^0 @xmath14 , @xmath15 and f_2(1275 ) @xmath18 rates .
however , this might well be due to overestimated fractions of the promptly produced a^+_0(980 ) @xmath22 and f_0(980 ) @xmath21 , which are difficult to estimate .
these presumably must be comparable with those for the vector mesons ( with similar masses ) , but are higher in jetset ( see table 1 ) .
such an explanation is supported by the mass dependence of the ^0 @xmath14 , @xmath15 , a^+_0(980 ) @xmath22 , f_0(980 ) @xmath21 and f_2(1275 ) @xmath18 total rates @xcite . if the production rates of other @xmath0-wave mesons follow the same mass dependence as observed in fig . 1
, this allows their production rates to be estimated .
for example , the corresponding predictions for the @xmath58 and @xmath59 total production rates per @xmath13 hadronic decay are @xmath60 and @xmath61 if the @xmath59 is a pure @xmath62 state .
moreover , provided that the observed mass dependence of production rates is indeed universal for all flavours , it allows not only the production rates of mesons with different flavours to be related , but also their masses .
indeed , from simple mass rescaling in fig . 1 one obtains the following phenomenological mass formulae : @xmath63_i = _
2^ * @xmath44- ( _ 2^ * @xmath44- ^ * @xmath25 ) t - p_i , @xmath64_i = ^*_2 @xmath65- ( ^*_2 @xmath65- ^ * @xmath66 ) t - p_i where @xmath67 , @xmath68 and @xmath69 are the masses of the vector , tensor and @xmath0-wave ( with @xmath55 or @xmath70 ) light - flavour mesons corresponding to the masses of their respective charm ^ * @xmath66 , ^*_2 @xmath65 and @xmath71 , and bottom ^ * @xmath25 , _ 2^ * @xmath44 and @xmath72 partners . from eq .
( 11 ) , with the ^*0 @xmath73 , _ 2^*0 @xmath74 , _ 1(1402 ) @xmath75 , ^ * @xmath66 , ^*_2 @xmath65 and ^ * @xmath25 masses from pdg @xcite and the _ 2^ * @xmath44 mass , @xmath76 mev/@xmath77 , taken as the average of the masses obtained by aleph , l3 and opal ( table 2 ) and with the error equal to half of the difference between the aleph and l3 values , one obtains : m__1 @xmath78 = 5728 16 /c^2 , m _
@xmath64 ^ 0_1 = 2433 6 /c^2 . with the _
1(1273 ) @xmath79 instead of the _ 1(1402 ) @xmath75 in eq .
( 11 ) one has : m_^*_1 @xmath45 = 5625 16 /c^2 , m _
@xmath64^*0_1 = 2325 6 /c^2 .
the limited accuracy of the phenomenological formulae ( 11 ) results in additional systematic uncertainty , not accounted for in the mass estimates given in eqs . ( 12 ) and ( 13 ) .
it can roughly be estimated from the mass relation @xmath80 following from eq . ( 11 ) , which imposes practically the same mass splitting between the _
2^ * @xmath44and _ 1 @xmath78 as between the ^*_2 @xmath65 and @xmath81 .
this is not consistent with the smaller _
2^ * @xmath44 and _ 1 @xmath78 mass difference of 12 mev/@xmath77 , required in the fits performed by the lep experiments , in comparison with the measured ^*0_2 @xmath82 and @xmath83 mass difference @xmath84 mev/@xmath77 and may result in possible biases of @xmath85 mev/@xmath77 in our mass estimates .
the _ 1 @xmath78 mass given in eq .
( 12 ) agrees within errors with the averaged value of this mass @xmath86 mev/@xmath77 from aleph , l3 and opal and @xmath87 mev/@xmath77 from cdf .
the mass difference @xmath88 mev/@xmath77 is consistent with the constraint of 12 mev/@xmath77 imposed by the lep experiments .
the ^*_1 @xmath45 mass given in eq .
( 13 ) is consistent within 2 standard deviations with @xmath89 mev/@xmath77 from the l3 fit , agrees with @xmath90 mev/@xmath77 from the aleph fit , but significantly smaller than the value following from the opal fit .
the obtained @xmath83 mass is in good agreement with the pdg value @xmath91 mev/@xmath77 .
the @xmath92 mass in eq .
( 13 ) represents our prediction for the mass of the broad , not yet established state .
the physical _ 1(1273 ) @xmath79 and _ 1(1402 ) @xmath75 are mixtures of the two su(3 ) octet states @xmath93 and @xmath94 .
the decay patterns of the _
1(1273 ) @xmath79 and _ 1(1402 ) @xmath75 suggest that these singlet and octet states are almost degenerate , with a mixing angle near @xmath95 .
thus , from the decay amplitudes of the _ 1(1273 ) @xmath79 and _ 1(1402 ) @xmath75 into @xmath96 and @xmath97 , the accmor collaboration found @xmath98 @xcite . provided that the heavy quark limit is also appropriate for the strange mesons ,
the two mixed _
1 @xmath99 mass eigenstates of @xmath100 can also be described by the total angular momentum @xmath101 of the light quark with @xmath51 and @xmath57 expected to be degenerate with the @xmath52 and @xmath102 states , respectively . by a change of basis
one can introduce a new mixing angle @xmath103 which defines the amount of @xmath51 and @xmath57 in the physical _ 1(1273 ) @xmath79 and _ 1(1402 ) @xmath75 states . according to isgur @xcite
( see also @xcite ) , the _
1(1273 ) @xmath79 and _ 1(1402 ) @xmath75 are quite near to being the pure @xmath57 and @xmath51 states , respectively .
the small splitting between ^*_0(1430 ) @xmath104 and _ 1(1402 ) @xmath75 , implying that the @xmath6 and @xmath4 levels are nearly degenerate as expected for all heavy - light systems , is well consistent with such association .
however , this is certainly not a case for the _ 2^*(1430 ) @xmath24 and _ 1(1273 ) @xmath79 associated with the @xmath7 and @xmath5 levels , respectively .
on the other hand , our results following from the mass relations ( 11 ) suggest that the _ 1(1402 ) @xmath75 and _ 1(1273 ) @xmath79 might be assigned in the heavy quark limit to the @xmath57 and @xmath51 levels , respectively .
this may certainly imply either that our assumption about a universal mass dependence of the production rates for all @xmath0-wave states ( including the _
1(1402 ) @xmath75 and _ 1(1273 ) @xmath79 ) fails , or that our phenomenological mass formulae ( 11 ) resulting from this assumption are not correct .
however , if it is not the case , such an assignment implies that the @xmath57 levels corresponding to the _ 1(1402 ) @xmath75 and _ 2^*(1430 ) @xmath24 are degenerate , whereas the @xmath51 levels corresponding to the _ 1(1273 ) @xmath79 and ^*_0(1430 ) @xmath104 are quite different , just contrary to the situation discussed earlier .
we also notice that an attempt to apply eq .
( 11 ) with the ^*_0(1430 ) @xmath104 mass for the determination of the _ 0 @xmath43 and @xmath105 masses results , due to the small difference between the _ 2^*(1430 ) @xmath24 and ^*_0(1430 ) @xmath104 masses , in @xmath106 and @xmath107 .
this is not consistent with the values of the ^*_1 @xmath45 and @xmath92 masses given in eq .
( 13 ) , if the mass difference between the ^*_1 @xmath45 and _ 0 @xmath43 , and also the @xmath92 and @xmath105 , is small as expected . for the _ 0 @xmath43 and _ 0 @xmath105 masses equal to the ^*_1 @xmath45 and @xmath92 masses given in eq .
( 13 ) , eq .
( 11 ) by definition gives the @xmath108 mass equal to the _
1(1273 ) @xmath79 mass .
a smaller @xmath108 mass is also required in the description of the light - flavour @xmath0-wave mesons in the nonrelativistic quark model @xcite . as noticed in @xcite ,
this can be explained if the observable ^*_0(1430 ) @xmath104 mass is replaced by its `` bare '' @xmath10 mass corresponding to the @xmath11-matrix pole . in the @xmath11-matrix analysis of the @xmath109-wave @xcite ,
the `` bare '' @xmath108 mass , in one of the two possible solutions , is @xmath110 mev/@xmath77 , consistent with the _
1(1273 ) @xmath79 mass .
thus , if this conjecture is correct , the _ 1(1402 ) @xmath75 and _ 1(1273 ) @xmath79 assignment in the heavy quark limit to the @xmath57 and @xmath51 levels is consistent with the expected degeneracy of the @xmath5 and @xmath7 and , respectively , the @xmath4 and @xmath6 levels .
it also provides a consistent description of the strange , charm and bottom meson production rates and also their masses and lends support to the models suggesting that the @xmath51 levels for the strange , charm and bottom mesons are _ below _ the @xmath111 levels . in particular , our results given in eq .
( 13 ) are in excellent agreement with the predictions @xcite .
in conclusion we have shown that the mass dependences of the production rates for the six families of primary produced mesons in @xmath13 hadronic decays obtained from results of the lep experiments : the vector and tensor light - flavour mesons , the vector and @xmath0-wave charm , strange charm and bottom mesons are very similar .
this allows not only the production rates of mesons with different flavours to be related , but also their masses , thus showing an interesting connection between hadron production properties and their masses .
our analysis suggests that the @xmath6 and @xmath4 levels are _ below _ the @xmath5 and @xmath7 levels not only for the charm and bottom but also for the strange mesons .
contrary to the conventional picture , the strange axial mesons _
1(1273 ) @xmath79 and _ 1(1402 ) @xmath75 might be considered as mainly @xmath4 and @xmath5 levels , respectively , degenerate with the @xmath6 and @xmath7 levels of the ^*_0(1430 ) @xmath104 and _ 2^*(1430 ) @xmath24 if the observed ^*_0(1430 ) @xmath104 mass is replaced by its `` bare '' @xmath10 mass corresponding to the @xmath11-matrix pole and close to the _ 1(1273 ) @xmath79 mass . although these results , if verified by future experiments , do not support the spin - orbit inversion suggested by isgur @xcite , they amusingly lend strong support to his conclusion about the key role that the strange quark plays as the link between heavy- and light - quark hadrons . |
the quantum mechanical description of neutrino oscillations @xcite has been the subject of much discussion and debate in the recent literature .
the ` standard ' oscillation formula @xcite , yielding an oscillation phase or @xmath1 . ] , at distance @xmath2 from the neutrino source , between neutrinos , of mass @xmath3 and @xmath4 and momentum @xmath5 , of : are used throughout . ]
@xmath6 is derived on the assumption of equal momentum at production of the two neutrino mass eigenstates .
other authors have proposed , instead , equal energies @xcite or velocities @xcite at production , confirming , in both cases , the result of the standard formula .
the latter reference claims , however , that the standard expression for @xmath7 should be multiplied by a factor of two in the case of the equal energy or equal momentum hypotheses
. however , the equal momentum , energy or velocity assumptions are all incompatible with energy - momentum conservation in the neutrino production process @xcite .
in two recent calculations @xcite a covariant formalism was used in which exact energy - momentum conservation was imposed . using the invariant feynman propagator @xcite to describe the space - time evolution of the neutrino mass eigenstates , the author of ref .
@xcite also found an oscillation phase a factor of two larger than given by the the standard formula .
however , on convoluting the transition amplitude with a gaussian wavefunction descibing the spatial - temporal position of the source pion , the standard result was recovered . in ref .
@xcite , a completely different formula was obtained for the neutrino oscillation phase following pion decay and it was predicted that only correlated oscillations of neutrinos and the recoiling decay muons could be observed .
however , the author of ref .
@xcite as well as others @xcite claimed that muon oscillations would either be completely suppressed , or essentially impossible to observe .
the present paper calculates the probabilities of oscillation of neutrinos and muons produced by pions decaying both at rest and in flight , as well as the probabilities of neutrino oscillation following muon decay or @xmath8decay of a nucleus at rest .
the calculations , which are fully covariant , are based on feynman s reformulation of quantum mechanics @xcite in terms of interfering amplitudes associated with classical space - time particle trajectories .
the essential interpretational formula of this approach , though motivated by the seminal paper of dirac on the lagrangian formulation of quantum mechanics @xcite , and much developed later in the work of feynman and other authors @xcite , was actually already given by heisenberg in 1930 @xcite .
the application of the path amplitude formalism to neutrino or muon oscillations is particularly staightforward , since , in the covariant formulation of quantum mechanics , energy and momentum are exactly conserved at all vertices and due to the macroscopic propagation distances of the observed neutrinos and muons all these particles follow essentially classical trajectories ( _ i.e. _ corresponding to the minima of the classical action ) which are rectilinear paths with constant velocities .
the essential formula of feynman s version of quantum mechanics , to be employed in the calculations presented below , is @xcite : @xmath9 where @xmath10 is the probability to observe a final state @xmath11 , given an initial state @xmath12 , and @xmath13 are ( unobserved ) intermediate quantum states . in the applications to be described in this paper , eqn(1.2 )
specialises to : @xmath14 for the case of neutrino oscillations and @xmath15 for the case of muon oscillations .
@xmath16 is the probability to observe the charged current neutrino interaction : @xmath17 following the decay : @xmath18 , while @xmath19 is the probability to observe the decay @xmath20 , after the same decay process . in eqns.(1.3),(1.4 ) @xmath21 are neutrino mass eigenstates while @xmath22 are the corresponding recoil muon states from pion decay .
@xmath23 , @xmath24 and @xmath25 denote invariant decay amplitudes , @xmath26 is the invariant amplitude of the charged current neutrino interaction , @xmath27 is the invariant space - time propagator of particle @xmath28 between the space - time points @xmath29 and @xmath30 and @xmath31 is an invariant amplitude describing the production of the @xmath32 by the source @xmath33 and its space - time propagation to the space - time point @xmath34 .
the difference of the approach used in the present paper to previous calculations presented in the literature can be seen immediately on inspection of eqns(1.3 ) and ( 1.4 ) .
the initial state contributes only a multiplicative constant to the transition probabilites .
the initial state can then just as well be defined as ` pion at @xmath34 ' , rather than @xmath35 .
this is done in the calculations presented in section 2 below . ]
is a pion at space time - point @xmath34 , the final state an @xmath36 or @xmath37 produced at space - time point @xmath38 .
these are unique points , for any given event and do not depend in any way on the masses of the unobserved neutrino eigenstates propagating from @xmath39 to @xmath38 in eqn(1.3 ) . on the other hand the ( unobserved ) space - time points @xmath39 at which the neutrinos and muons are produced _ do _ depend on @xmath40 . indeed , because of the different velocities of the propagating neutrino eigenstates , only in this case can both neutrinos and muons ( representing _ alternative _ classical histories of the decaying pion ) both arrive simultanously at the unique point @xmath38 where the neutrino interaction occurs ( eqn(1.3 ) ) or the muon decays ( eqn(1.4 ) ) .
the crucial point in the above discussion is that the decaying pion , _ via _ the different path amplitudes in eqns(1.3 ) and ( 1.4 ) , _ interferes with itself_. to modify very slightly dirac s famous statement : ` each pion then interferes only with itself .
interference between two different pions never occurs ' . as will be seen in the discussion in section 5 below ,
essentially all recent treatments of neutrino oscillations in the literature have attempted to contradict the second sentence in dirac s statement ! because of the different possible decay times of the pion in the two interfering path amplitudes , the pion propagators @xmath41 in eqns(1.3 ) and ( 1.4 ) above give important contributions to the interference phase . to the author s best knowledge ,
this effect has not been taken into account in any previously published calculation of neutrino oscillations .
the results found for the oscillation phase are , for pion decays at rest : @xmath42 and for pion decays in flight : @xmath43 where @xmath44 the superscripts indicate the particles whose propagators contribute to the interference phase .
also @xmath45 , @xmath46 and @xmath47 are the energies of the parent @xmath48 and the decay @xmath49 and @xmath50 and @xmath51 , @xmath52 the angles between the pion and the neutrino , muon flight directions . in eqns.(1.5 ) to ( 1.7 ) terms of order @xmath53 , @xmath54 , and higher , are neglected , and in eqns.(1.6 ) and ( 1.7 ) ultrarelativistic kinematics with @xmath55 is assumed .
formulae for the oscillation phase of neutrino oscillations following muon decays or nuclear @xmath0-decays at rest , calculated in a similar manner to eqn(1.5 ) , are given in section 3 below .
a brief comment is now made on the generality and the covariant nature of the calculations presented in this paper .
although the fundamental formula ( 1.2 ) is valid in both relativistic and non - relativistic quantum mechanics , it was developed in detail by feynman @xcite only for the non - relativistic case .
for the conditions of the calculations performed in the present paper ( propagation of particles in free space ) the invariant space - time propagator can either be derived ( for fermions ) from the dirac equation , as originally done by feynman @xcite or , more generally , from the covariant feynman path integral for an arbitary massive particle , as recently done in ref .
@xcite . in the latter case
, the invariant propagator for any stable particle with pole mass @xmath56 , between space - time points @xmath57 and @xmath58 in free space is given by the path integral @xcite : @xmath59 \exp \left \
{ -\frac{im}{2 } \int_{x_i}^{x_f } \left ( \frac{dx}{d\tau } \cdot \frac{dx}{d\tau}+1 \right)d \tau\right \}\ ] ] where @xmath60 is the proper time of the particle . by splitting the integral over @xmath61 on the right side of eqn(1.8 ) into the product of a series of infinitesimal amplitudes corresponding to small segments , @xmath62 , gaussian integration
may be performed over the intermediate space - time points . finally , integrating over the proper time @xmath60 , the analytical form of the propagator
is found to be @xcite : @xmath63 where : @xmath64 and @xmath65 is a first order hankel function of the second kind , in agreement with ref .
@xcite . in the asymptotic region where @xmath66 , or for the propagation of on - shell particles @xcite
, the hankel function reduces to an exponential and yields the configuration space propagator @xmath67 of eqn(2.11 ) below .
it is also shown in ref .
@xcite that energy and momentum is exactly conserved in the interactions and decays of all such ` asymptotically propagating ' particles .
the use of quasi - classical particle trajectories and the requirement of exact energy - momentum conservation are crucial ingredients of the calculations presented below .
the structure of the paper is as follows . in the following section the case of neutrino or muon oscillations following pion decay at rest
is treated .
full account is taken of the momentum wave - packets of the progagating neutrinos and muons resulting from the breit - wigner amplitudes describing the distributions of the physical masses of the decaying pion and daughter muon .
the corresponding oscillation damping corrections and phase shifts are found to be very small , indicating that the quasi - classical ( constant velocity ) approximation used to describe the neutrino and muon trajectories is a very good one .
the incoherent effects , of random thermal motion of the source pion , and of finite source and detector sizes , on the oscillation probabilities and the oscillation phases , are also calculated .
these corrections are found to be small in typical experiments , but much larger than those generated by the coherent momentum wave packets . in section 3 , formulae are derived to describe neutrino oscillations following muon decay at rest or the @xmath0-decay of radioacive nuclei . these are written down by direct analogy with those derived in the previous section for pion decay at rest . in section 4 ,
the case of neutrino and muon oscillations following pion decay in flight is treated . in this case
the two - dimensional spatial geometry of the particle trajectories must be related to the decay kinematics of the production process . due to the non - applicability of the ultrarelativistic approximation to the kinematics of the muon in the pion rest frame , the calculation , although straightforward , is rather tedious and lengthy for the case of muon oscillations , so the details are relegated to an appendix .
finally , in section 5 , the positive aspects and shortcomings of previous treatments in the literature of the quantum mechanics of neutrino and muon oscillations are discussed in comparison with the method and results of the present paper .
to understand clearly the different physical hypotheses and approximations underlying the calculation of the particle oscillation effects it is convenient to analyse a precise experiment .
this ideal experiment is , however , very similar to lnsd @xcite and karmen @xcite except that neutrinos are produced from pion , rather than muon , decay at rest .
the different space - time events that must be considered in order to construct the probability amplitudes for the case of neutrino oscillations following pion decay at rest are shown in fig.1 .
a @xmath32 passes through the counter c@xmath68 , where the time @xmath69 is recorded , and comes to rest in a thin stopping target t ( fig.1a ) ) . for simplicity ,
the case of only two neutrino mass eigenstates @xmath70 and @xmath71 of masses @xmath3 and @xmath4 ( @xmath3 @xmath72 @xmath4 ) is considered .
the pion at rest constitutes the initial state of the quantum mechanical probability amplitudes .
the final state is an @xmath73 system produced , at time @xmath74 , via the process @xmath75 at a distance @xmath2 from the decaying @xmath32 ( fig.1d ) ) .
two different physical processes may produce the observed @xmath73 final state , as shown in fig.1b ) and 1c ) , where the pion decays either at time @xmath76 into @xmath70 or at time @xmath77 into @xmath71 .
the probability amplitudes for these processes are , up to an arbitary normalisation constant : @xmath78 note that following the conventional ` @xmath79 ' ( final , initial ) ordering of the indices of matrix elements in quantum mechanics , the path amplitude is written from right to left in order of increasing time .
this ensures also correct matching of ` bra ' and ` ket ' symbols in the amplitudes . in eqn(2.1 ) , @xmath80 , @xmath81 are the invariant amplitudes of the @xmath82 charged current scattering and pion decay processes , @xmath83 , @xmath84 and @xmath85 , @xmath86 are relativistic breit - wigner amplitudes , @xmath87 and @xmath88 are amplitudes describing the mixture of the flavour and mass neutrino eigenstates , and @xmath89 is the lorentz - invariant configuration space propagator @xcite of the pion or neutrino .
the pole masses and total decay widths of the pion and muon are denoted by @xmath90 , @xmath91 and @xmath92 , @xmath93 respectively . for simplicity ,
phase space factors accounting for different observed final states are omitted in eqn(2.1 ) and subsequent formulae . because the amplitudes and propagators in eqn(2.1 ) are calculated using relativistic quantum field theory , and the neutrinos propagate over macroscopic distances , it is a good approximation , as already discussed in the previous section , to assume exact energy - momentum conservation in the pion decay process , and that the neutrinos are on their mass shells ,
_ i.e. _ @xmath94 , where @xmath95 is the neutrino energy - momentum four - vector . in these circumstances
the neutrino propagators correspond to classical , rectilinear , particle trajectories .
the pion and muon are unstable particles whose physical masses @xmath96 and @xmath97 differ from the pole masses @xmath98 and @xmath92 appearing in the breit - wigner amplitudes and covariant space - time propagators in eqn(2.1 ) .
the neutrino momentum @xmath99 will depend on these physical masses according to the relation : @xmath100[w_{\pi}^2-(m_i - w_{\mu(i)})^2 ] \right]^{\frac{1}{2 } } } { 2 w_{\pi}}\ ] ] note that , because the initial state pion is the same in the two path amplitudes in eqn(2.1 ) @xmath96 does not depend on the neutrino mass index @xmath12 .
however , since the pion decays resulting in the production of @xmath70 and @xmath71 are independent physical processes , the physical masses of the unobserved muons , @xmath101 , recoiling against the two neutrino mass eigenstates are not , in general , the same . in the following kinematical calculations sufficient accuracy
is achieved by retaining only quadratic terms in the neutrino masses , and terms linear in the small quantities : @xmath102 , @xmath103 .
this allows simplification of the relativistic breit - wigner amplitudes : @xmath104 developing eqn(2.2 ) up to first order in @xmath105 , @xmath106 and @xmath107 yields the relation : @xmath108 \end{aligned}\ ] ] where @xmath109 the term @xmath110 which is also included in eqn(2.4 ) gives a negligible o(@xmath111 ) contribution to the neutrino oscillation formula . for muon oscillations ,
however , it gives a term of o(@xmath105 ) in the interference term , as discussed below .
similarly , the exact formula for the neutrino energy : @xmath112 in combination with eqn(2.4 ) gives , for the neutrino velocity : @xmath113~+~o(m_i^4,\delta_{\pi}^2,\delta_i^2)\ ] ] this formula will be used below to calculate the neutrino times - of - flight : @xmath114 .
the unitary transformation specifying the mixing of the flavour and mass neutrino eigenstates is defined by a single real mixing angle @xmath115 : @xmath116 and the mass eigenstates @xmath117 respect the orthonormality conditions : @xmath118 the parts of the amplitudes requiring the most careful discussion are the invariant space - time propagators @xmath89 , as it is mainly their treatment that leads to the different result for the neutrino oscillation phase found in the present paper , as compared to those having previously appeared in the literature . in the limit of large time - like separations
, the propagator may be written as @xcite : @xmath119\ ] ] @xmath89 is the amplitude for a particle , originally at a space - time point ( @xmath120 , @xmath121 ) , to be found at ( @xmath122 , @xmath123 ) and @xmath124 , @xmath125 . in the following , according to the geometry of the experiment shown in fig.1 , only one spatial coordinate will be considered ( @xmath126 ) and only the exponential factor in eqn(2.11 ) , containing the essential phase information for particle oscillations will be retained in the amplitudes .
solid angle correction factors , taken into account by the factor in large brackets in eqn(2.11 ) , are here neglected , but are easily included in the final oscillation formulae .
writing then @xmath127 = \exp[-im\delta \tau ] \equiv \exp[-i\delta \phi]\ ] ] it can be seen that the increment in phase of the propagator , @xmath128 , when the particle undergoes the space - time displacement ( @xmath129 , @xmath130 ) is a lorentz invariant quantity equal to the product of the particle mass and the increment , @xmath62 , of proper time .
using the relativistic time dilatation formula : @xmath131 and also the relation , corresponding to a classical , rectilinear , particle trajectory : @xmath132 gives , for the phase increments corresponding to the paths of the neutrinos and the pion in fig.1 : @xmath133 \\
\delta \phi_i^{\pi } & = & m_{\pi } ( t_i - t_0 ) = m_{\pi } ( t_d - t_0)-\frac{m_{\pi}l}{v_i } \nonumber \\ & = & m_{\pi } ( t_d - t_0)-m_{\pi}l\left\{1+\frac{m_i^2}{2 p_0 ^ 2}\left[1- \frac{2\delta_{\pi}}{m_{\pi}}\frac{(m_{\pi}^2+m_{\mu}^2)}{(m_{\pi}^2-m_{\mu}^2 ) } + \frac{4\delta_i m_{\mu}}{m_{\pi}^2-m_{\mu}^2}\right]\right\ } \end{aligned}\ ] ] to perform the integral over @xmath106 in eqn(2.17 ) it is convenient to approximate the modulus squared of the breit - wigner amplitude by a gaussian , _ via _ the substitution : @xmath140 where the width of the gaussian is chosen so that it has approximately the same full width at half maximum , @xmath141 , as the breit - wigner function . after the substitution ( 2.22 ) , the integral over @xmath106 in eqn(2.17 ) is easily evaluated by a change of variable to ` complete the square ' in the argument of the exponential , with the result : @xmath142\ ] ] since @xmath143 mev , the term dependent on @xmath91 in eqn(2.19 ) may be neglected , so that @xmath144 may be taken as a real number with the numerical value : @xmath145 for typical physically interesting values ( see below ) of @xmath146 ev and @xmath147 m , @xmath144 takes the value 0.48 mev@xmath148 , so that @xmath149 then , to very good accuracy , @xmath150 , independently of the neutrino mass .
it follows that for neutrino oscillations , the muon mass dependence of the amplitudes may be neglected for any physically interesting values of @xmath151 and @xmath2 . from eqns ( 2.17 ) and ( 2.23 ) the probability to observe
the reaction @xmath75 at distance @xmath2 from the pion decay point and at time @xmath74 is : @xmath152\delta_{\pi}\right]\right\ } \nonumber\end{aligned}\ ] ] the time dependent exponential factors in the curly brackets of eqn(2.25 ) are easily understood . if @xmath153 then @xmath154 .
this implies that the neutrino of mass @xmath3 results from an earlier decay than the neutrino of mass @xmath4 , in order to be detected at the same time .
because of the exponential decrease with time of the pion decay amplitude , the contribution to the probability of the squared amplitude for the neutrino of mass @xmath3 is larger . the interference term resulting from the product of the decay amplitudes of the two neutrinos of different mass , has an exponential factor that is the harmonic mean of those of the squared amplitudes for each neutrino mass eigenstate , and so is also suppressed relative to the squared amplitude for the neutrino of mass @xmath3 .
the integral over the physical pion mass is readily performed by replacing the breit - wigner function by a gaussian as in eqn(2.22 ) .
this leads to an overall multiplicative constant @xmath155 and a factor : @xmath156\ ] ] multiplying the interference term . for @xmath157 and @xmath147
m the numerical value of this factor is @xmath158 .
this tiny correction is neglected in the following equations .
integrating over @xmath74 gives the average probability to observe the @xmath73 final state at distance @xmath2 : @xmath159 \cos\frac{2 m_{\pi } m_{\mu}^2 \delta m^2 } { ( m_{\pi}^2- m_{\mu}^2)^2}l\right\ } \end{aligned}\ ] ] where all kinematical quantities are expressed in terms of @xmath160 , @xmath90 and @xmath92 .
note that the minimum value of @xmath74 is @xmath161 , @xmath162 and @xmath161 for the squared amplitude terms of neutrinos of mass @xmath3 , @xmath4 and the interference term , respectively . on integrating over @xmath74 , the squared amplitude terms give equal contributions , the larger amplitude for mass @xmath3 being exactly compensated by a smaller range of integration .
the exponential damping factor in the interference term in eqn(2.26 ) is derived using the relations : @xmath163 and @xmath164 to obtain @xmath165 the damping factor arises because the difference in the times - of - flight of the two neutrino paths is limited by the pion lifetime .
it will be seen below , however , that for distances @xmath2 of practical interest for the observation of neutrino oscillations , the damping effect is tiny .
the part of the oscillation phase in eqn(2.25 ) originating from the neutrino propagators ( the term associated with the ` 1 ' within the large curved brackets ) differs by a factor two from the corresponding expression in the standard formula .
the contribution to the oscillation phase of the propagator of the decaying pion ( the term associated with @xmath166 in the large curved brackets of eqn(2.24 ) ) has not been taken into account in any published calculation known to the author of the present paper . the oscillation phase in eqn(2.26 )
is @xmath167 times larger than that given by the standard formula ( 1.1 ) . for @xmath168 m , as in the lnsd experiment , the first oscillation maximum occurs for @xmath169 . denoting by @xmath170 the phase of the cosine interference term in eqn(2.26 ) , the pion lifetime damping factor can be written as : @xmath171 so the damping effect is vanishingly small when @xmath172 .
the oscillation formula ( 2.26 ) is calculated on the assumption that the decaying pion is at rest at the precisely defined position @xmath57 .
in fact , the positive pion does not bind with the atoms of the target , but will rather undergo random thermal motion .
this has three effects : an uncertainy in the value of @xmath57 , a doppler shift of the neutrino energy and a time dilatation correction correction factor of @xmath173 in the equation ( 2.16 ) for the pion phase increment . assuming that the target is at room temperature ( t= @xmath174 k ) , the mean kinetic energy of @xmath175 correponds to a mean pion momentum of @xmath176 mev and a mean velocity of @xmath177 km / sec .
the pion will move , in one mean lifetime ( @xmath178 sec ) , a distance of 146@xmath179 .
this is negligible as compared to @xmath2 ( typically @xmath180 30 m ) and so eqn(2.26 ) requires no modification to account for this effect .
the correction factor due to the doppler effect and time dilatation is readily calculated on the assumption of a maxwell - boltzmann distribution of the pion momentum : @xmath181 here @xmath182 mev .
details of the calculation are given in appendix a. the interference term in eqn(2.26 ) is modified by a damping factor : @xmath183l\right)^2 \right\}\ ] ] while the argument of the cosine term acquires an additional phase factor : @xmath184l\ ] ] for @xmath185 , @xmath186 and @xmath187 .
if the target in which the pion stops is of thickness @xmath188 , then the effect of different stopping points of the @xmath48 ( assumed uniformly distributed ) is to multiply the interference term in ( 2.26 ) by the factor : @xmath189 if the position of the neutrino interaction point within the target has an uncertainy of @xmath190 a similar correction factor is found , with the replacement @xmath191 in eqn(2.34 ) .
the calculation of this correction factor is also described in appendix a. if the target t in which the pion comes to rest ( fig.1a ) ) is chosen to be sufficiently thin , the pion decay process may be detected by observing the recoil muon in the counter c@xmath68 at times @xmath76 or @xmath77 ( fig.1b ) or 1c ) ) .
a sufficiently accurate measurement of the times @xmath76 , @xmath77 and @xmath74 would , in principle , enable separation of the different processes in figs.1b ) and 1c ) by the observation of separated peaks in the time - of - flight distribution at @xmath192 and @xmath193 . in this case
the interference term in eqn(2.26 ) vanishes as the two alternative space - time paths leading to the neutrino interaction are distinguishable .
however , for @xmath168 m and @xmath194(ev)@xmath195 the difference in the times of flight is only @xmath196 sec , more than ten orders of magnitude smaller than can be measured with existing techniques . as discussed in section 5 below , the momentum smearing due to the doppler effect at room temperature is some eleven orders of magnitude larger than the shift due to a neutrino mass difference with @xmath194(ev)@xmath195 .
thus , even with infinitely good time resolution , separation of such neutrino mass eigenstates by time - of - flight is not possible . the ideal experiment , described above to study neutrino oscillations , is easily adapted to the case of oscillations in the decay probability of muons produced by charged pion decay at rest . as previously pointed out in ref .
@xcite , such oscillations will occur if neutrinos with different masses exist .
as before , the pion stops in the target t at time @xmath69 ( see fig.2a ) ) . at time
@xmath76 the pion decays into @xmath70 and the corresponding recoil muon ( @xmath197 ) , whose passage is recorded in the counter c@xmath198 ( fig.2b ) ) .
similarly , a decay into @xmath71 and @xmath199 may occur at time @xmath77 ( fig.2c ) ) . with a suitable choice of the times @xmath76 and @xmath77 , such that muons following the alternative paths both arrive at the same time @xmath74 at the point @xmath58 , interference occurs between the path amplitudes when muon decay occurs at the space - time point ( @xmath58 , @xmath74 ) in the detector d ( fig.2d ) ) .
the path amplitudes corresponding to muons recoiling against neutrinos of mass @xmath3 and @xmath4 are : @xmath200 the various factors in these equations are defined , _
mutadis mutandi _ , as in eqn(2.1 ) . with the same approximations , concerning the neutrino masses and the physical pion and muon masses , as those made above
, the velocity of the muon recoiling against the neutrino mass eigenstate @xmath201 is : @xmath202\end{aligned}\ ] ] where @xmath203 comparing with eqn(2.7 ) , it can be seen that for the muon case , unlike that where neutrino interactions are observed , there are pion and muon mass dependent correction terms that are independent of the neutrino masses , implying a velocity smearing effect due to the physical pion and muon masses that is @xmath204 larger than for the case of neutrino oscillations .
the phase increments corresponding to the paths of the muons and the pion in fig .
2 are , using ( 2.4 ) . ] and ( 2.12)-(2.14 ) : @xmath205 \end{aligned}\ ] ] @xmath206 @xmath207 @xmath208\ ] ] where @xmath209 using eqns(2.12),(2.38 ) and ( 2.39 ) to re - write the space - time propagators in eqn(2.35 ) , as well as eqn(2.3 ) for the breit - wigner amplitudes gives : @xmath210 d \delta_i~~i=1,2 \end{aligned}\ ] ] where @xmath211 @xmath212\ ] ] @xmath213\ ] ] and @xmath214 making the substitution ( 2.22 ) and performing the integral over @xmath106 according to eqn(2.23 ) , the following formula is found for the probability for muon decay at distance @xmath2 and time @xmath74 .
@xmath215 ^ 2 |<e^+\nu_e\overline{\nu}_{\mu}|t|\mu^+>|^2 e^{-\frac{\gamma_{\mu}v_0^{\mu}}{\gamma_0^{\mu}}l } \nonumber \\ & & |<\nu_{\mu } \mu^+|t|\pi^+>|^2 e^{-\gamma_{\pi}(t_d - t_0 ) } \frac{\gamma_{\pi}^2}{4(\delta_{\pi}^2+\frac{\gamma_{\pi}^2}{4 } ) } \nonumber \\ & & \left\{\sin^2\theta e^{\gamma_{\pi } t_{\mu(1)}^f } + \cos^2 \theta e^{\gamma_{\pi } t_{\mu(2)}^f } \right .
\nonumber \\ & & -2\sin \theta \cos \theta e^{\frac{\gamma_{\pi}}{2}(t_{\mu(1)}^f + t_{\mu(2)}^f ) } \nonumber \\ & & \left .
{ \it re } \exp i \left [ \frac{m_{\mu}^2 \delta m^2}{2 p_0
^ 3}\left(1-\frac { e_0^{\mu}}{m_{\pi}}\right)l + [ \alpha_{\pi}^{\mu}(1)-\alpha_{\pi}^{\mu}(2)]\delta_{\pi } \right ] \right\}\end{aligned}\ ] ] the muon path difference yields the term associated with @xmath216 in the interference phase in eqn(2.46 ) while the pion path is associated with ` 1 ' in the large round brackets .
the numerical value of the damping factor : @xmath217 [ 1-\frac{\alpha^{\mu } \gamma_{\mu}}{3}]^2\ ] ] resulting from the integral over the physical muon mass is , for @xmath147 m , 0.774 , so , unlike for the case of neutrino oscillations , the correction is by no means negligible .
this is because , in the muon oscillation case , the leading term of @xmath218 is not proportional to the neutrino mass squared .
the non - leading terms proportional to @xmath105 have been neglected in eqn(2.43 ) .
this correction however effects only the overall normalisation of the oscillation formula , not the functional dependence on @xmath2 arising from the interference term . integrating over @xmath107 using eqns(2.22 ) and ( 2.23 ) , as well as over @xmath74 , gives the probability of muon decay at distance @xmath2 from the production point , where all kinematical quantities are expressed in terms of @xmath160 , @xmath90 and @xmath92 : @xmath219[1- \frac{4}{3 } \frac{\gamma_{\mu } m_{\mu } m_{\pi } } { m_{\pi}^2- m_{\mu}^2}l]^2 \nonumber \\ & & |<e^+\nu_e\overline{\nu}_{\mu}|t|\mu^+>|^2 \exp \left[-\frac{2 \gamma_{\mu } m_{\pi } m_{\mu } ( m_{\pi}^2- m_{\mu}^2)}{(m_{\pi}^2 + m_{\mu}^2)^3 } l \right]|<\nu_{\mu } \mu^+|t|\pi^+>|^2 \nonumber \\ & & \left\{1-\sin 2 \theta\exp[-\frac{2 \gamma_{\pi } m_{\pi}^2 m_{\mu}^2 \delta m^2 } { ( m_{\pi}^2- m_{\mu}^2)^3}l ] \cos \frac{2 m_{\pi } m_{\mu}^2 \delta m^2 } { ( m_{\pi}^2- m_{\mu}^2)^2 } l \right\}\end{aligned}\ ] ] in this expression the correction due to the damping factor of the interference term : @xmath220\ ] ] arising from the integral over the physical pion mass has been neglected . for @xmath157 and @xmath147 m the numerical value of this factor is @xmath221 . denoting by @xmath222 the argument of the cosine in eqn(2.47 ) , the exponential damping factor due to the pion lifetime may be written as : @xmath223 for @xmath224 , @xmath225 so , as in the neutrino oscillation case , the pion lifetime damping of the interference term is very small .
corrections due to time dilatation and the doppler effect are calculated in a similar way to the neutrino oscillation case with the results ( see appendix a ) : @xmath226l\right)^2 \right\}\ ] ] and @xmath227l\ ] ] as for neutrino oscillations , the corresponding corrections are very small for oscillation phases of order unity .
the phase of the cosine in the interference term is the same in neutrino and muon oscillations , as can be seen by comparing eqns(2.26 ) and ( 2.47 ) .
it follows that the target or detector size correction ( eqn(2.34 ) ) is the same in both cases .
neutrino and muon oscillations from pion decay at rest have an identical oscillation phase for given values of @xmath160 and @xmath2 . in view of the much larger event rate that is possible , it is clearly very advantageous in this case to observe muons rather than neutrinos .
in fact , it it not necessary to observe muon decay , as in the example discussed above .
the oscillation formula applies equally well if the muons are observed at the distance @xmath2 using any high efficiency detector such as , for example , a scintillation counter . in contrast
, the rate of neutrino oscillation events is severely limited by the very small neutrino interaction cross section .
the formula describing neutrino oscillations @xmath228 following the decay at rest of a @xmath229 , @xmath20 is easily derived from the similar formula where a @xmath230 is produced by @xmath32 decay at rest ( 2.25 ) . because the neutrino momentum spectrum is continous , smearing effects due to the finite muon lifetime
may be neglected from the outset .
the phase increment associated with the neutrino path is then given by eqn(2.15 ) with the replacements @xmath231 and @xmath232 , where @xmath233 is the antineutrino momentum .
the phase increment of the decaying muon is given by the same replacements in eqn(2.16 ) and , in addition , @xmath234 and @xmath235 .
the formula , analogous to eqn(2.26 ) , for the time - averaged probability to detect the process @xmath236 at a distance l from the muon decay point is then : @xmath237 \cos \left[\frac{\delta m^2 } { p_{\overline{\nu}}}\left ( \frac{m_{\mu}}{2 p_{\overline{\nu}}}-1\right)l \right ] \right\}\end{aligned}\ ] ] the standard neutrino oscillation formula , hitherto used in the analysis of all experiments , has instead the expression @xmath238 for the argument of the cosine term in eqn(3.1 ) . denoting my @xmath239
the value of @xmath160 obtained using the standard formula , and @xmath240 that obtained using the feynman path ( fp ) formula ( 3.1 ) then : @xmath241 for a typical value of @xmath242 of 45 mev , eqn(3.2 ) implies that @xmath243 .
thus the @xmath244 oscillation signal from @xmath229 decays at rest reported by the lnsd collaboration @xcite corresponding to @xmath245 ( ev)@xmath195 for @xmath246 implies @xmath247 ( ev)@xmath195 for a similar mixing angle . for the case of @xmath0-decay : @xmath248 @xmath90 in the first line of eqn(2.16 ) is replaced by @xmath249 , the total energy release in the @xmath0-decay process : @xmath250 where @xmath251 and @xmath252 are the masses of the parent and daughter nuclei . that the phase advance of an unstable state , over a time , @xmath130 , is given by @xmath253 where @xmath254 is the excitation energy of the state , is readily shown by the application of time - dependent perturbation theory to the schrdinger equation @xcite .
a more intuitive derivation of this result has been given in ref .
@xcite . in the present case , @xmath255 .
omitting the lifetime damping correction , which is about eight orders of magnitude smaller than for pion decay , given a typical @xmath0-decay lifetime of a few seconds , the time - averaged probabilty to detect the @xmath256 via the process @xmath257 , at distance @xmath2 from the decay point is given by the formula , derived in a similar way to eqns(2.26 ) and ( 3.1 ) : @xmath258 \right\}\end{aligned}\ ] ] where @xmath259 the lifetime of the unstable nucleus @xmath260 .
until now , all experiments have used the standard expression @xmath238 for the neutrino oscillation phase .
the values of @xmath160 found should be scaled by the factor @xmath261 , suitably averaged over @xmath242 , to obtain the @xmath160 given by the feynman path formula ( 3.4 ) .
in this section , the decays in flight of a @xmath32 beam with energy @xmath262 into @xmath263 are considered .
as the analysis of the effects of the physical pion and muon masses have been shown above to give negligible corrections to the @xmath2 dependence of the oscillation formulae , for the case of decays at rest , such effects will be neglected in this discussion of in - flight decays .
the overall structure of the path amplitudes for neutrinos and muons is the same as for decays at rest ( see eqns(2.1 ) and ( 2.35 ) ) .
however , for in - flight decays , in order to calculate the interfering paths originating at different and terminating at common space - time points , the two - dimensional spatial geometry of the problem must be properly taken into account . in fig.3
a pion decays at a into the 1 mass eigenstate , the neutrino being emitted at an angle @xmath264 in the lab system relative to the pion flight direction . if @xmath265 a later pion decay into the 2 mass eigenstate at the angle @xmath266 may give a path such that both eigenstates arrive at the point b at the same time .
a neutrino interaction @xmath267 , or @xmath268 occuring at this space - time point will than be sensitive to interference between amplitudes corresponding to the paths ab and acb .
the geometry of the triangle abc and the condition that the 1 and 2 neutrino mass eigenstates arrive at b at the same time gives the following condition on their velocities : @xmath269 expanding to first order in the small quantity @xmath270 , rearranging and neglecting terms of @xmath271 gives : @xmath272\ ] ] where the relation : @xmath273 has been used .
rearranging eqn(4.2 ) : @xmath274\ ] ] the difference in phase of the neutrino paths ab and cb is ( see eqn(2.15 ) ) : @xmath275 since the angle @xmath276 is @xmath277 , the difference between ab and cd is of the same order , and so : @xmath278 where @xmath279 , the measured neutrino energy . from the geometry of the triangle abc
: @xmath280 so , to first order in @xmath276 , and using eqn(4.4 ) : @xmath281 and @xmath282 eqns(4.8 ) and ( 4.9 ) give , for the phase increment of the pion path : @xmath283 in eqn(4.10 ) , the lorentz invariant character of the propagator phase is used . setting @xmath284 and @xmath285 , the pion phase increment of eqn(2.25 )
is recovered .
eqns(4.6 ) and ( 4.10 ) give , for the total phase difference of the paths ab , acb : @xmath286\ ] ] using the expressions , valid in the ultra - relativistic ( ur ) limit where @xmath287 : @xmath288 and @xmath289 where @xmath290 is the angle between the directions of the pion and neutrino momentum vectors in the pion rest frame , eqn(4.11 ) may be rewritten as : @xmath291 for @xmath292 the oscillation phase is the same as for pion decay at rest ( see eqn(2.26 ) ) since in the latter case , @xmath293 . using eqn(4.14 ) , the probability to observe a neutrino interaction , at point b , produced by the decay product of a pion decay occuring within a region of length , @xmath294 , ( @xmath295 ) centered at the point a , in a beam of energy @xmath45 , is given by a formula analagous to eqn(2.26 ) : @xmath296 as in the case of pion decay at rest , eqn(2.26 ) , the oscillation phase differs by the factor @xmath167 from that given by the standard formula .
the derivation of the formula describing muon oscillations following pion decays in flight is very similar to that just given for neutrino oscillations .
the condition on the velocities so that the muons recoiling against the different neutrino mass eigenstates arrive at the point b ( see fig.3 ) at the same time , is given by a formula analagous to ( 4.2 ) : @xmath297 } { v_{\pi}\sin\theta_1}\delta\ ] ] the formula relating the neutrino masses to the muon velocities is , however , more difficult to derive than the corresponding relation for neutrinos , ( 4.3 ) , as the decay muons are not ultra - relativistic in the pion rest frame .
the details of this calculation are given in appendix b. the result is : @xmath298 using eqn(4.16 ) and the relation , valid to first order in @xmath276 , : @xmath299 where @xmath130 is the flight time of the pion from a to c in fig.3 ( and also the difference in the times - of - flight of the muons recoiling against the two neutrino eigenstates ) , the angle @xmath276 may be eliminated to yield : @xmath300}\ ] ] using now the kinematical relation ( see appendix b ) : @xmath301 and the expression for the phase difference of the paths ab and acb : @xmath302 together with eqn(4.19 ) , it is found , taking the ur limit , where @xmath303 , that @xmath304\ ] ] the probability of detecting a muon decay at b is then : @xmath305 |<\nu_{\mu } \mu^+|t|\pi^+>|^2 \nonumber \\ & & \times \left\{1-\sin 2 \theta \cos \frac { 2 m_{\mu}^2 \delta m^2}{e_{\mu}^2 ( m_{\pi}^2-m_{\mu}^2)^2 ) } \left[\frac{(m_{\mu}^2 e_{\pi}- m_{\pi}^2 e_{\mu})l}{\cos\theta_1}\right ] \right\ } \end{aligned}\ ] ] where @xmath294 is defined in the same way as in eqn(4.13 ) .
the quantum mechanics of neutrino oscillations has been surveyed in two recent review articles @xcite , where further extensive lists of references may be found .
the standard derivation of the neutrino oscillation phase will first be considered , following the treatment of ref .
@xcite , but using the notation of the present paper .
the calculation is performed in terms of the ` flavour eigenstates of the neutrino ' , @xmath306 , which in the case of pion decay is , at @xmath307 , essentially the pure state @xmath308 corresponding to the superposition of the mass eigenstates @xmath309 , @xmath310 given in eqn(2.9 ) above .
this flavour eigenstate is assumed to evolve with laboratoy time , @xmath311 , according to fixed energy solutions of the non - relativistic schrdinger equation : @xmath312 where @xmath313 are mass eigenstates of fixed momentum @xmath314 , and @xmath315 , @xmath316 are the laboratory energies of the neutrino mass eigenstates .
the amplitude for transition into the state @xmath317 at time @xmath311 is then , using eqns(2.8 ) , ( 2.9 ) : @xmath318 assuming now that the neutrinos have the same momentum but different energies : @xmath319 and using ( 5.2 ) and ( 5.3 ) , the probability of the flavour state @xmath82 at time @xmath311 is found to be : @xmath320 \right)\ ] ] finally , since the velocity difference of the neutrino mass eigenstates is o(@xmath160 ) , then , to the same order in the oscillation phase , the replacement @xmath321 can be made in eqn(5.4 ) to yield the standard oscillation phase of eqn(1.1 ) . *
the time evolution of the neutrino mass eigenstates in eqn(5.1 ) according to the schrdinger equation yields a non - lorentz - invariant phase @xmath322 , to be compared with the lorentz - invariant phase @xmath323 given in eqn(2.12 ) above .
although the two expressions agree in the non - relativistic limit @xmath324 it is clearly inappropriate to use this limit for the description of neutrino oscillation experiments .
it may be noted that the lorentz - invariant phase is robust relative to different kinematical approximations .
the same result is obtained to order @xmath325 for the phase of spatial oscillations independent of whether the neutrinos are assumed to have equal momenta or energies .
this is not true in the non - relativistic limit . assuming equal momenta gives the standard result of eqn(1.1 ) , whereas the equal energy hypothesis results in a vanishing oscillation phase .
a contrast may be noted here with the standard treatment of neutral kaon oscillations , which follows closely the derivation in eqns(5.1 ) to ( 5.4 ) above , except that the particle phases are assumed to evolve with time according to @xmath326 $ ] where @xmath56 is the particle mass and @xmath60 is its proper time , in agreement with eqn(2.12 ) . * as pointed out in ref .
@xcite , the different neutrino mass eigenstates do not have equal momenta as assumed in eqns(5.1 ) and ( 5.3 ) .
the approximation of assuming equal momenta might be justified if the fractional change in the momentum of the neutrino due to a non - vanishing mass were much less than that of the energy . however , in the case of pion decay as is readily shown from eqns(2.4 ) and ( 2.6 ) above , the ratio of the fractional shift in momentum to that in energy is actually @xmath327 ; so , in fact , the opposite is the case . * to derive ( 5.2 ) from ( 5.1 ) , it is necessary to introduce the concept of a ` flavour momentum eigenstate ' which can not be defined in a theoretically consistent manner @xcite . * what are the physical meanings of @xmath311 , @xmath314 in eqn(5.1 ) ? in eqn(5.2 ) , it is assumed that the neutrino mass eigenstates are both produced , and both detected , at the same times .
thus both have the same time - of - flight @xmath311 .
the momentum @xmath314 can not be the same for both eigenstates , as assumed in eqn(5.3 ) , if both energy and momentum are conserved in the decay process .
for any given value of the laboratory time @xmath311 the different neutrino mass eigenstates must be at different space - time positions because they have different velocities , if it is assumed that both mass eigenstates are produced at the same time .
it then follows that the @xmath82 part of the different mass eigenstates can not be probed at , some space - time point , by a neutrino interaction , whereas the latter must clearly occur at some unique and specific point in every event .
* the historical development of the calculation of the neutrino oscillation phase is of some interest .
the first published prediction @xcite actually obtained a phase a factor two larger than eqn(1.1 ) _ i.e. _ in agreement with the covariant calculation of ref.@xcite(before the introduction of ` wave packets ' ) and with the contribution from neutrino propagation found in the present paper .
this prediction was later used , for example , in ref.@xcite .
the derivation sketched above , leading to the standard result of eqn(1.1 ) was later given in ref.@xcite
. a subsequent paper @xcite by the authors of ref.@xcite , published shortly afterwards , cited both ref.@xcite and ref.@xcite , but used now the prediction of the latter paper .
no comment was made on the factor of two difference in the two calculations . in a later review article @xcite by the authors of
ref.@xcite a calculation similiar to that of ref.@xcite was presented in detail .
subsequently , all neutrino oscillation experiments have been analysed on the assumption of the standard oscillation phase of eqn(1.1 ) .
it may be thought that the kinematical and geometrical inconsistencies mentioned in points ( ii ) and ( iv ) above result from a too classical approach to the problem .
after all , what does it mean , in quantum mechanics , to talk about the ` position ' and ` velocity ' of a particle , in view of the heisenberg uncertainty relations @xcite ?
this point will become clear later in the present discussion , but first , following the original suggestion of ref .
@xcite , and , as done in almost all subsequent work on the quantum mechanics of neutrino oscillations , the ` wave packet ' description of the neutrino mass eigenstates will be considered . in this approach , both the ` source ' and also possibly the ` detector ' in the neutrino oscillation experiment are described by coherent spatial wave packets . here
the ` source ' wavepacket treatment in the covariant approach of ref .
@xcite will be briefly sketched . after discussing the results obtained , and comparing them with those of the present paper , the general consistency of the wave packet approach with the fundamental quantum mechanical formula ( 1.2 ) will be examined .
the basic idea of the wave packet approach of ref .
@xcite is to replace the neutrino propagator ( 2.11 ) in the path amplitude by a four - dimensional convolution of the propagator with a ` source wavefunction ' which , presumably , describes the space - time position of the decaying pion . for mathematical convenience
this wavefunction is taken to have a gaussian form with spatial and temporal widths @xmath328 and @xmath329 respectively .
thus , the neutrino propagator @xmath89 is replaced by @xmath330 where : @xmath331 where @xmath57 and @xmath58 are 4-vectors and @xmath332\ ] ] the integral in ( 5.5 ) was performed by the stationary phase method , yielding the result ( up to multiplicative and particle flux factors ) , and here assuming , for simplicity , one dimensional spatial geometry : @xmath333\ ] ] for the case of @xmath334 oscillations , following @xmath32 decay at rest , the probability to observe flavour @xmath82 at time @xmath311 and distance @xmath335 is given by : @xmath336 performing the integral over @xmath130 and making the ultra - relativistic approximation @xmath337 yields finally , with @xmath338 : it can be seen that the oscillation phase in eqn(5.9 ) is the same as standard one of eqn(1.1 ) .
the exponential damping factors in eqn(5.9 ) are the same as those originally found in ref .
@xcite for spatial wave packets ( @xmath340 ) .
considering now only spatial wave packets and using the property @xmath341 derived from the fourier transform of a gaussian , the two terms in the exponential damping factor may be written as : @xmath342\ ] ] and @xmath343\ ] ] the spatial damping factor @xmath344 is usually interpreted in terms of a ` coherence length'@xcite . if @xmath345 then @xmath346 and the neutrino oscillation term is strongly suppressed .
eqn(5.12 ) expresses the condition that oscillations are only observed provided that the wave packets overlap . since @xmath347 the separation of the wave packets is @xmath348 , so that eqn(5.12 ) is equivalent to @xmath349 ( no wave packet overlap ) .
the damping factor @xmath350 is typically interpreted @xcite in terms of the ` heisenberg uncertainty principle ' .
this factor is small , unless the difference in mass of the eigenstates is much less than @xmath351 , so it is argued that only for wide momentum wave packets can neutrino oscillations be observed , whereas in the contrary case , when the mass eigenstates are distinguishable , the interference effect vanishes . in the case of pion decay
the difference in momentum of the two interfering mass eigenstates comes only from the @xmath106 term in eqn(2.4 ) , as @xmath107 , being a property of the common initial state , is the same for both eigenstates .
the neutrino momentum smearing in pion decay is then estimated from eqn(2.4 ) as : @xmath352 for @xmath353 the value of @xmath350 is found to be @xmath354 giving complete suppression of neutrino oscillations .
this prediction is in clear contradiction with the tiny damping corrections found in the path amplitude analysis in of section 2 above .
the convolution formula ( 5.5 ) may be re - written so as to isolate the contribution from the region : @xmath355 where @xmath356 is an arbitary fixed value of the coordinate @xmath29 ( say , the spatial position of source pion parallel to the direction of neutrino propagation ) : here the @xmath30 , @xmath358 , @xmath359 averaged quantity @xmath360 is a function of @xmath361 , @xmath362 as well as @xmath363 and @xmath364 .
similarly : @xmath365 now in calculating the oscillation probability ( 5.8 ) , eqns(5.13 ) and ( 5.14 ) will give a contribution to the interference term from the terms proportional to @xmath366 , @xmath367 of : @xmath368 this contribution corresponds to interference between a pion decaying at an arbitary point @xmath369 with another decaying at another arbitary point @xmath370 .
this is in contradiction with eqn(1.2 ) , according to which , _ the initial quantum state must be the same for all interfering path amplitudes_. referring to fig 1 .
, it can be seen that the different path amplitudes correspond to two ( classically alternative ) decay histories of the _ same pion _ which must clearly be at some unique spatial position at time @xmath69 . `
each pion interferes only with itself .
interference between two _ different _ pions never occurs ' @xcite .
thus the convolution , at amplitude level , of the neutrino propagator with ` source ' and/or ` detector ' wavefunctions does not respect heisenberg and feynman s fundamental law of quantum mechanics ( 1.2 ) .
the preceding discussion of the derivation of the standard formula for the oscillation phase ( 1.1 ) in terms of ` flavour eigenstates ' revealed contradictions and inconsistencies if the neutrinos are assumed to follow classical space - time trajectories .
the ` source wave packet ' treatment gives the standard result for the oscillation phase and predicts that the interference term will be more or less damped depending on the widths in configuration and momentum space of the wave packets .
however the convolution formula ( 5.5 ) is evidently at variance with eqn(1.2 ) and the requirement that particles can interfere only ` with themselves ' .
so do wave packets actually play a role in the correct quantum mechanical description of neutrino oscillations , as suggested in ref.@xcite ?
are the packets actually constrained by the heisenberg uncertainty relations ?
how do the properties of the wave packet effect the possiblity to observe neutrino oscillations ? the answers to all these questions are contained in the results of the calculations presented in section 2 above .
they are now reviewed , with special emphasis on the basic assumptions made and the physical interpretation of the equations . referring again to fig.1 , in a ) a single pion comes to rest in the stopping target t. the time of its passage is recorded by the counter c@xmath68 , which thus defines the initial state as a @xmath32 at rest at time @xmath69 .
this pion , being an unstable particle , has a mass @xmath96 that is , in general , different from its most likely value which is the pole mass @xmath90 .
what are shown in fig.1b ) and fig.1c ) are two different classical histories of this _ very same pion_. in b ) it decays into the neutrino mass eigenstate @xmath70 at time @xmath76 , and in c ) it decays into the neutrino mass eigenstate @xmath71 at time @xmath77 . because these are independent classical histories , the physical masses @xmath371
, @xmath372 of the muons recoiling against the mass eigenstates @xmath70 and @xmath71 respectively , are , in general , not equal .
taking into account now exact energy - momentum conservation in the decay processes ( appropriate because of the covariant formulation used throughout ) the eigenstates @xmath70 and @xmath71 will have momenta which depend on @xmath96 and @xmath371 and @xmath96 and @xmath372 respectively .
these momenta are calculated in eqn(2.4 ) .
the distributions of @xmath96 and @xmath97 are determined by breit - wigner amplitudes , that are the fourier transforms of the corresponding exponential decay laws , in terms of the decay widths @xmath91 and @xmath93 respectively . in accordance with eqn(1.2 ) , only the breit - wigner amplitudes corresponding to the physical muon masses are ( coherently ) integrated over at the amplitude level . the integral over @xmath96 ( a property of the initial state ) is performed ( incoherently ) at the level of the transition probability . because of the long lifetimes of the @xmath48 and @xmath50 , the corresponding momentum wave packets are very narrow , so that all corrections resulting from integration over the resulting momentum spectra are found to give vanishingly small corrections . indeed ,
as is evident from the discussion of eqn(5.10 ) above , the width of the momentum wave packet is much smaller than the difference in the momenta of the eigenstates expected for experimentally interesting values of the neutrino mass difference ( say , @xmath373 ) .
however , contrary to the prediction of eqn(5.10 ) , this does not at all prevent the observation of neutrino oscillations .
this is because the oscillations result from interference between amplitudes corresponding to different propagation times , not momenta , of the neutrino mass eigenstates .
table 1 shows contributions to the fractional smearing of the neutrino momentum from different sources : * neutrino mass difference @xmath373 ( eqn(2.4 ) ) * coherent effect due to the physical muon mass ( @xmath374 in eqn(2.4 ) ) * incoherent effect due to the physical pion mass ( @xmath375 in eqn(2.4 ) ) * incoherent doppler effect assuming @xmath376 the pion mass effect is of the same order of magnitude as the neutrino mass shift .
the muon mass effect is two orders of magnitude smaller , while the doppler effect at room temperature gives a shift eleven orders of magnitude larger than a ( 1ev)@xmath195 neutrino mass difference squared .
according to the usual interpretation of the heisenberg uncertainty principle , the neutrinos from pion decay , which , as has been shown above , correspond to very narrow momentum wave packets , would be expected to have a very large spatial uncertainty . indeed , interpreting the width of the coherent momentum wave packet generated by the spread in @xmath377 according to the the momentum - space uncertainty relation @xmath378 gives @xmath379 km .
does this accurately represent the knowledge of the position of a decay neutrino obtainable in the experiment shown in fig.1 ? without any experimental difficulty , the decay time of the pion can be measured with a precision of @xmath380sec , by detecting the decay muon .
thus , at any later time the distance of the neutrino from the decay point is known with a precision of @xmath381 cm .
this is a factor 4@xmath382 more precise than the ` uncertainty ' given by the heisenberg relation .
it is clear that the experimental knowledge obtainable on the position of the neutrino is essentially classical , in agreement with the theoretical description in terms of classical particle trajectories in space - time . in this case
the momentum - position uncertainty relation evidently does not reflect the possible experimental knowledge of the position and momentum of the neutrino .
this is because it does not take into account the prior knowledge that the mass of the neutrino is much less than that of the pion or muon , so that its velocity is , with negligible uncertainty , c. in spite of this , a heisenberg uncertainty relation is indeed respected in the pion decay process .
the breit - wigner amplitude that determines the coherent spread of neutrino momentum is just the fourier transform of the exponential decay law of the muon .
the width parameter of the breit - wigner amplitude and the muon mean lifetime do indeed respect the _ energy - time _
uncertainty relation @xmath383 .
it is then clear , from this careful analysis of neutrino oscillations following pion decay at rest , that , in contradiction to what has been almost universally assumed until now , _
the neutrinos are not described by a coherent spatial wave packet_. there is a coherent momentum wave packet , but it is only a kinematical consequence of a breit wigner amplitude . if for mathematical convenience
, the momentum wave packet is represented by a gaussian , a conjugate ( and spurious ) gaussian spatial wave packet will be generated by fourier transformation .
indeed , in the majority of wave packet treatments that have appeared in the literature , gaussian momentum and spatial wave packets related by a fourier transform with widths satisfying the ` uncertainty relation ' @xmath384 have been used .
the above discussion shows that such a treatment does not correspond to the actual knowledge of the space - time events that constitute realistic neutrino oscillation experiments . the limitation on the detection distance @xmath2 , for observation of neutrino oscillations , given by the damping factor @xmath385 of eqn(2.30 ) is easily understood in terms of the classical particle trajectories shown in fig.1 . for a given velocity difference , the time difference @xmath386 becomes very large when both neutrinos , in the alternative classical histories , are required to arrive simultaneously at a far distant detector . because of the finite pion lifetime , however , the amplitude for pion decay at time @xmath77 is smaller than that at @xmath76 by the factor @xmath387 $ ] . integrating over all decay times results in the exponential damping factor @xmath385 of eqn(2.30 ) .
it is clear that , contrary to the damping factor @xmath344 of eqn(5.11 ) , the physical origin of the @xmath2 dependent damping factor is quite unrelated to ` wavepacket overlap ' . for a given value of @xmath386 the coherent neutrino momentum spread originating in the breit wigner amplitude for @xmath377
produces a corresponding velocity smearing that reduces the number of possible classical trajectories arriving at the detection event .
this effect is taken into account in the integral shown in eqn(2.23 ) .
the effect is shown , in section 2 above , to be much smaller than the ( already tiny ) pion lifetime damping described above , and it is neglected in eqn(2.26 ) . finally , as descibed after eqn(2.25 ) , the ( incoherent ) integration over the breit wigner function containing @xmath96 gives an additional damping correction to the interference term that is also very tiny compared to that due to the pion lifetime .
as promised above , a final remark is now made on the physical interpretation of the damping factors ( 5.10 ) and ( 5.11 ) that have often been derived and discussed in the literature .
although as pointed out above , and to be further discussed below , the derivation of these factors is incorrect , as it results from coherent integration over different spatial positions of the neutrino source , similar factors do appear when performing the integrations over the ( unobserved ) neutrino momentum distributions .
@xmath344 is replaced by @xmath385 and @xmath350 by the factor resulting from the coherent integration over the physical muon mass @xmath377 .
the reason for the huge suppression factor predicted by eqn(5.10 ) , and the tiny one found in the path amplitude calculation , is that , in deriving ( 5.10 ) and ( 5.11 ) , it is assumed that the neutrino eigenstates are both produced and detected at equal times .
this will only be possible if both the hypothetical ` wave packet overlap ' is appreciable ( eqn(5.11 ) ) and also the momentum smearing is sufficiently large that the time - of - flight differences due to the different neutrino masses are washed out ( eqn(5.10 ) . in the path amplitude calculation interference and hence oscillations are made possible by different _ decay times _ of the source pion , and the damping factors analagous to @xmath344 and @xmath350 turn out to give vanishingly small corrections to the oscillation term .
it may be remarked that the physical interpretation of ` neutrino oscillations ' provided by the path amplitude description is different from the conventional one in terms of ` flavour eigenstates ' . in the latter the amplitudes of different flavours in the neutrino
are supposed to vary harmonically as a function of time . in the amplitudes for the different physical processes in the path amplitude treatment
there is , instead , no variation of lepton flavour in the propagating neutrinos .
if the mass eigenstates are represented as superpositions of flavour eigenstates using the inverses of eqns(2.8 ) and ( 2.9 ) , there is evidently no temporal variation of the lepton flavour composition . only in the detection process itself
are the flavour eigenstates projected out , and the interference effect occurs that is described as ` neutrino oscillations ' . in the case of the observation of the recoil muons
no such projection takes place , but exactly similar interference effects are predicted to occur . as previously emphasised @xcite , the ` flavour
oscillations ' of neutrinos , neutral kaons and b - mesons are just special examples of the universal phenomenon of quantum mechanical superposition that is the basis of eqn(1.2 ) . by far the most widespread difference from the path amplitude treatment of the present paper is the non - respect of the basic quantum mechanical formula , ( 1.2 ) , by the introduction of wave packets to describe ` source ' and/or ` detector ' particles[7,8,10,11,30 - 46 ] . since in any practical neutrino oscillation experiment a single initial or final quantum state , as specified by eqn(1.2 ) , is not defined , but rather sets of initial and final states @xmath388 and @xmath389 determined by experimental conditions , eqn(1.2 ) may be generalised to : @xmath390 to be contrasted with the formula used in the references cited above : @xmath391 here , @xmath392 and @xmath393 are ` source ' and ` detector ' wave packets respectively . in eqn(5.17 )
the initial states @xmath394 and final states @xmath395 of eqn(5.16 ) , that correspond to different spatial positions , and also , possibly , different kinematic properties , of the source particle or detection event , are convoluted with spatial and/or temporal ` wave packets ' , that , as discussed above , are devoid of physical significance .
use of eqn(5.17 ) leads to the absurd prediction ( see , for example , eqn(5.15 ) above ) that _ different _ initial or final state particles interfere with each other . a possible reason for the widespread use of eqn(5.17 ) instead of ( 5.16 )
may be understood following a remark of the author of ref .
@xcite concerning a ` paradox ' of the complete quantum field theory calculation that takes into account , by a single invariant amplitude , production , propagation and detection of the neutrinos .
it was noticed that , if the amplitude for the complete chain of processes is considered to correspond to one big feynman diagram , then integration over the space - time coordinates of the initial and final states will reduce the exponential factors , containing the essential information on the interference phase , to energy - momentum conserving delta functions , and so no oscillations will be possible .
a related remark was made by the authors of ref .
@xcite who stated that , as they were assuming exact energy - momentum conservation , the integration over the space - time coordinates could be omitted .
they still , however , ( quite inconsistently , in view of the previous remark ) retained the exponential factors containing the interference phase information .
these considerations indicate a general confusion between momentum space feynman diagram calculations , where it is indeed legitimate to integrate , at the amplitude level , over the unobserved space - time positions of the initial and final state particles , and the case of neutrino oscillations , where the amplitude must be defined in configuration space . in the latter case ,
it is the unobserved momenta of the propagating particles that should be integrated over , as is done in eqns(2.1 ) and ( 2.35 ) above , and not the space - time positions of the ` source ' or ` detector ' particles , as in eqn(5.17 ) .
it is clear that exact energy - momentum conservation plays a crucial role in the path amplitude calculation .
this is valid only in a fully covariant theory .
still , several authors , in spite of the ultra - relativistic nature of neutrinos , used a non - relativistic theory to describe the production , propagation and detection of neutrinos @xcite .
as is well known , in such ` old fashioned perturbation theory ' @xcite energy is not conserved at the level of propagators and so no precise analysis of the kinematics and the space - time configurations of the production and detection events , essential in the covariant path amplitude analysis , is possible .
even when , in some cases , the complete production , propagation and detection process of the neutrinos were described[35 - 37,39,40 ] , equal neutrino energies [ 37,39,40 ] , equal neutrino momenta [ 36 ] or either @xcite were assumed , in contradiction with energy - momentum conservation and a consistent space - time description of the production and detection events .
as follows directly from eqn(5.3 ) ( or the similar formula , for the neutrino momentum , obtained by assuming equal neutrino energies ) , in all the above cited references , the standard formula ( 1.1 ) for the oscillation phase was obtained .
an interesting discussion of the interplay between different kinematical assumptions ( not respecting energy - momentum conservation ) and the space - time description of the production and detection events was provided in ref .
this treatment was based on the lorentz - invariant propagator phase of eqn(2.12 ) . by assuming either equal momentum or equal energy for the propagating neutrinos , but allowing _ different times of propagation _ for the two mass eigenstates , values of @xmath396 agreeing with eqns(2.15 ) and ( 2.25 ) above
were found , _
i.e. _ differing by a factor two from the standard formula .
alteratively , assuming equal velocities , ( and hence equal propagation times ) the standard result ( 1.1 ) was obtained . in this latter case , however , the masses , momenta and energies of the neutrinos must be related , up to corrections of o(@xmath105 ) , according to : @xmath397 since the ratio of the neutrino masses may take , in general , any value , so must then the ratio of their momenta . for the case of neutrino production from pion decay at rest with @xmath398
the relation ( 5.18 ) is clearly incompatible with eqn(2.4 ) which gives : @xmath399 on the other hand , eqn(5.19 ) is clearly compatible ( up to corrections of o(@xmath160 ) ) with the ` equal momentum ' hypothesis .
there is similar compatiblity with the ` equal energy ' hypothesis .
even so , the authors of ref .
@xcite recommended the use of the equal velocity hypothesis . with the hindsight provided by the path amplitude analysis , in which the two neutrino mass eigenstates do indeed have different propagation times
, it can be seen that the kinematically consistent ` equal momentum ' and ` equal energy ' choices are good approximations and the neutrino oscillation phase , resulting from the propagation of the neutrinos alone , is indeed a factor two larger than the prediction of the standard formula . only the author of ref .
@xcite included the propagator of the decaying pion in the complete production - propagation - detection amplitude ; compare eqn(8 ) of ref .
@xcite with eqn(2.1 ) above . however , no detailed space - time analysis of production and detection events was performed .
square spatial wave packets for the ` source ' and ` detector ' were convoluted at amplitude level as in eqn(5.17 ) . as the neutrinos were assumed to have equal energies
the standard result for the oscillation phase was obtained . although , following ref .
@xcite , most recent studies of the quantum mechanics of neutrino oscillations have considered the complete production - propagation - detection process , some authors still use , in spite of the criticisms of ref .
@xcite , the ` flavour eigenstate ' description @xcite . in the last two of these references
a ` quantum field theory ' approach ia adopted , leading to an ` exact ' oscillation formula @xcite that does not make use of the usual ultra - relativistic approximation .
this work did not include either exact kinematics , or an analysis of the space - time structure of production and detection events .
also the effect of the propagator of the source particle was not considered . in ref .
@xcite the equal energy hypothesis was used and in ref .
@xcite the equal momentum hypothesis . as a consequence , in all three cases
the standard result was found for the oscillation phase in the ultra - relativistic limit .
correlated production and detection of neutrinos and muons produced in pion decay were considered in ref .
the introduction to this paper contains a valuable discussion of the universality of the ` particle oscillation ' phenomenon .
it is pointed out that this is a consequence of the general principle of amplitude superposition in quantum mechanics , and so is not a special property of the @xmath400 , @xmath401 and neutrino systems which are usually discussed in this context .
this paper used a covariant formalism that employed the ` energy representation ' of the space - time propagator . in the introduction , the important difference between eqns(5.16 ) and ( 5.17 ) was also touched upon : even so , in the amplitude for the correlated detection of the muon and neutrino ( eqn(2.10 ) of ref .
@xcite ) not only are the space - time positions of the production points of the neutrino and muon integrated over , but they are assumed to be at _ different _ space - time points .
the propagator of the decaying pion is not included in the amplitude , and although exact energy - momentum conservation is imposed , no space - time analysis of the production and decay points is performed .
correlated spatial oscillations of neutrinos and muons are predicted , though with interference phases different from the results of both the present paper and the standard formula . integrating over the spatial position of either the detected neutrino or muon destroys the interference term , so it is predicted that only correlated neutrino - muon oscillations may be observed , following pion decay , in contradiction with the results of the present paper , and , ( in the case of neutrino oscillations ) with all previous studies .
pion and muon lifetime effects were mentioned in ref .
@xcite , but neither the role of the pion lifetime in enabling different propagation times for the neutrinos nor the momentum smearing , induced by the fourier - transform - related breit wigner amplitudes , were discussed . the claim of ref .
@xcite ) that correlated neutrino - muon oscillations should be observable in pion decay was questioned in ref .
the authors of the latter paper attempted to draw conclusions on the possibility , or otherwise , of particle oscillations by using ` plane waves ' , _ i.e. _ energy - momentum eigenfunctions .
as is well known , such wavefunctions are not square integrable , and so can yield no spatial information . the probability to find a particle described by such a wave function in any finite spatial volume is zero .
due to the omission of the ( infinite ) normalisation constants of the wavefunctions many of the equations in ref .
@xcite are , as previously pointed out @xcite , dimensionally incorrect .
momentum wavepackets for the decaying pion convoluted at amplitude level as in eqn(5.17 ) were also discussed in ref . @xcite .
although exact energy - momentum conservation constraints were used , it was assumed , as in ref .
@xcite , that the muons and the different neutrino mass eigenstates are both produced and detected at common points ( eqn(35 ) of ref .
the latter assumption implies equal velocities , yielding the standard neutrino oscillation phase as well as the inconsistent kinematical relation ( 5.18 ) .
the authors of ref .
@xcite concluded that : * correlated @xmath402 oscillations of the type discussed in ref .
@xcite could be observed , though with different oscillation phases .
* oscillations would not be observed if only the muon is detected * neutrino oscillations can be observed even if the muon is not detected .
conclusion ( b ) is a correct consequence of the ( incorrect ) assumption that the muons recoiling against the different neutrino mass eigenstates have the same velocity . as both muons have the same mass they will have equal proper time increments .
so according to eqn(2.12 ) the phase increments will also be equal and the interference term will vanish .
the conclusion ( c ) is in contradiction to the prediction of eqn(3.22 ) of ref .
@xcite , according to which , no neutrino oscillations can be observed when the decay position of muon is integrated over .
the path amplitude calculation of the present paper shows that conclusion ( b ) is no longer valid when the different possible times of propagation of the recoiling muons are taken into account .
observation of neutrino oscillations following pion decay , using a covariant formalism ( schwinger s parametric integral representation of the space - time propagator ) was considered in ref .
exact energy - momentum conservation was imposed , but integration over the pion spatial position at amplitude level , as in eqn(5.17 ) , was done and no account was taken of the contribution of the pion space - time propagator to the oscillation phase . as the different eigenstates were assumed to be be produced or detected at identical space - time points equal propagation velocities were implicitly assumed , so that just as in refs@xcite , where the same assumption was made , the standard neutrino oscillation phase was obtained . in the conclusion of this paper the almost classical nature of the space - time trajectories followed by the neutrinos was stressed , although this was not taken to its logical conclusion in the previous discussion , _
e.g. _ the kinematical inconsistency of the equal velocity hypothesis that requires the evidently impossible condition ( 5.18 ) to be satisfied . in a recent paper @xcite , the standard neutrino oscillation formula with oscillation phase given by eqn(1.1 )
was compared with a neutrino decoherence model . in order to take into account incertainties in the position of the source and the neutrino energy , an average was made over the quantity @xmath403 , assuming that it is distributed according to a gaussian with mean value @xmath404 and width @xmath405 .
the average was performed in an incoherent manner .
thus the calculation is closely analagous to those for the effects of target or detector length or of thermal motion of the neutrino source , presented in the appendix a of the present paper . perhaps uniquely then , in the published literature , in ref .
@xcite the effects of source position and motion are taken into account correctly , according to eqn(5.16 ) instead of eqn(5.17 ) .
however , the source of the neutrino energy uncertainty is not specified . in
as far as it is generated from source motion the calculation is , in principle , correct .
there is however also the ( typically much smaller , see table 1 above , for the case of pion decay at rest ) coherent contribution originating from the variation in the physical masses of the unstable particles produced in association with the neutrino , as discussed in detail above .
it was concluded in ref .
@xcite that the gaussian averaging procedure used gave equivalent results to the decoherence model for a suitable choice of parameters .
it is clearly of great interest to apply the calculational method developed in the present paper to the case of neutral kaon and b - meson oscillations . indeed
the use of the invariant path amplitude formalism has previously been recommended @xcite for experiments involving correlated pairs of neutral kaons . here , just a few remarks will be made on the main differences to be expected from the case of neutrino or muon oscillations .
a more detailed treatment will be presented elsewhere@xcite . in the case of neutrino and muon oscillations ,
the interference effect is possible as the different neutrino eigenstates can be produced at different times .
this is because the decay lifetimes of all interesting sources ( pions , muons , @xmath0-decaying nuclei ) are much longer than the time difference beween the paths corresponding to the interfering amplitudes . to see
if a similar situation holds in the case of @xmath406 oscillations , three specific examples will be considered with widely differing momenta of the neutral kaons : these correspond to neutral kaon momenta of 108 mev , 750 mev and 5 gev respectively . in each case
the time difference ( @xmath411 ) of production of @xmath412 and @xmath413 mesons , in order that they arrive at the same time at a point distant @xmath414 ( where @xmath415 is the usual relativistic parameter ) from the source is calculated .
exact relativistic kinematics is assumed and only leading terms in the mass difference @xmath416 are retained . taking the value of @xmath417 and the various particle masses from ref .
@xcite the following results are found for @xmath411 in the three cases : ( i ) 3.4@xmath418sec , ( ii ) 8.3@xmath419sec and ( iii ) 6.4@xmath420sec . for comparison , for neutrino oscillations following pion decay at rest , with @xmath421 and
@xmath422 m , eqn(2.29 ) gives @xmath423sec . the result ( i )
may be compared with the mean life of the @xmath424 meson of 1.5@xmath425sec @xcite .
thus the @xmath424 lifetime is a factor of about five larger than @xmath411 indicating that @xmath406 interference should be possible by a similar mechanism to neutrino oscillations following pion decays , _
i.e. _ without invoking velocity smearing of the neutral kaon mass eigenstates . in cases ( i ) and
( ii ) the interference effects observed will depend on the ` characteristic time ' of the ( non resonant ) strong interaction process , a quantity that has hitherto not been susceptible to experimental investigation .
if this time is much less than , or comparable to , @xmath411 , essentially equal velocities ( and therefore appreciable velocity smearing ) of the eigenstates will be necessary for interference to occur .
since @xmath417 and @xmath426 are comparable in size , velocity smearing effects are expected to be , in any case , much larger than for neutrino oscillations following pion decay .
these effects are readily calculated using the gaussian approximation ( 2.22 ) of the present paper .
the main contribution to the velocity smearing is due to the variation of the physical mass of the @xmath412 rather than those of the @xmath413 or @xmath427 . for the @xmath428 oscillation case , analagous to ( i ) above
, @xmath429 ( @xmath430 mev ) the value of @xmath431 is found to be @xmath432sec , to be compared with @xmath433sec @xcite , which is a factor 3.8 smaller .
thus , velocity smearing effects are expected to play an important role in @xmath428 oscillations .
this is possible , since the neutral b - meson decay width ( @xmath434 mev ) , and mass difference ( @xmath435 mev ) , have similar sizes . in closing ,
it is interesting to mention two types of atomic physics experiments where interference effects similar to the conjectured ( and perhaps observed @xcite ) neutrino oscillations have aleady been clearly seen .
the first is quantum beat spectroscopy @xcite .
this type of experiment , which has previously been discussed in connection with neutrino oscillations @xcite , corresponds closely to the gedanken experiment used by heisenberg @xcite to exemplify the the fundamental law of quantum mechanics eqn(1.2 ) .
the atoms of an atomic beam are excited by passage through a thin foil or a laser beam .
the quantum phase of an atom with excitation energy @xmath254 evolves with time according to : @xmath436 ( see the discussion after eqn(3.3 ) above ) .
if decay photons from two nearby states with excitation energies @xmath437 and @xmath438 are detected after a time interval @xmath130 ( for example by placing a photon detector beside the beam at a variable distance @xmath439 from the excitation foil ) a cosine interference term with phase : @xmath440 where @xmath441 is the average velocity of the atoms in the beam , is observed @xcite .
an atom in the beam , before excitation , corresponds to the neutrino source pion .
the excitation process corresponds to the decay of the pion .
the propagation of the two different excited states , _ alternative _ histories of the initial atom , correspond to the _ alternative _ propagation of the two neutrino mass eigenstates .
finally the dexcitation of the atoms and the detection of a _ single _ photon corresponds to the neutrino detection process .
the particular importance of this experiment for the path amplitude calculations presented in the present paper , is that it demonstrates , experimentally , the important contribution to the interference phase of the space - time propagators of excited atoms , in direct analogy to the similar contributions of unstable pions , muons and nuclei discussed above . an even closer analogy to neutrino osillations following pion decay is provided by the recently observed process of photodetachement of an electron by laser excitation : the ` photodetachment microscope ' @xcite .
a laser photon ejects the electron from , for example , an @xmath442 ion in a beam .
the photodetached electron is emitted in an s - wave ( isotropically ) and with a fixed initial energy .
it then moves in a constant , vertical , electric field that is perpendicular to the direction of the ion beam and almost parallel to the laser beam .
an upward moving electron that is decelerated by the field eventually undergoes ` reflection ' before being accelerated towards a planar position - sensitive electron detector situated below the beam and perpendicular to the electric field direction ( see fig.1 of ref .
@xcite ) . in these circumstances
, it can be shown @xcite that , just two classical electron trajectories link the production point to any point in the kinematically allowed region of the detection plane .
typical parameters for @xmath442 are @xcite : initial electron kinetic energy , 102 kev ; detector distance , 51.4 cm ; average time - of - flight , 117 ns ; difference in emission times to arrive in spatial - temporal coincidence at the detector plane , 160 ps .
an interference pattern is generated by the phase difference between the amplitudes corresponding to the two allowed trajectories .
the phase difference , derived by performing the feynman path integral of the classical action along the classical trajectories @xcite , gives a very good description of the observed interference pattern .
the extremely close analogy between this experiment and the neutrino oscillation experiments described in sections 2 and 3 above is evident .
notice that the neutrinos , like the electrons in the photodetachement experiment , must be emitted at different times , in the alternative paths , for interference to be possible .
this is the crucial point that was not understood in previous treatments of the quantum mechanics of neutrino oscillations .
actually , ref .
@xcite contains , in section iv , a path amplitude calculation for electrons in free space that is geometrically identical to the discussion of pion decays in flight presented section 4 above ( compare fig.3 of the present paper with fig.3 of ref .
@xcite ) the conclusion of ref .
@xcite is that , in this case , no interference effects are possible for electrons that are mononergetic in the source rest frame . as is shown in section 4 above ,
if these electrons are replaced either by neutrinos of different masses from pion decay , or muons recoiling against such neutrinos , observable interference effects are indeed to be expected .
random thermal motion of the decaying pion in the target has two distinct physical effects on the phase of neutrino oscillations , @xmath443 ( where the first and second terms in eqn(a1 ) give the contributions of the neutrino and pion paths respectively ) : * the observed neutrino momentum , @xmath444 , is no longer equal to @xmath445 , due to the boost from the pion rest frame to the laboratory system .
( doppler effect or lorentz boost ) * the time increment of the pion path @xmath446 ( see eqn(2.16 ) ) no longer corresponds to the pion proper time .
( relativistic time dilatation ) taking into account ( 1 ) and ( 2 ) gives , for the neutrino oscillation phase : @xmath447 where : @xmath448 here , @xmath449 is the angle between the neutrino momentum vector and the pion flight direction in the pion rest frame .
developing @xmath450 and @xmath451 in terms of the small quantity @xmath452 , eqn(a2 ) may be written as : @xmath453 \cos\theta_{\nu}^{\ast } + \left(\frac{p_{\pi}}{m_{\pi}}\right)^2 \frac{\delta m^2 l}{2 p_0}\left[1-\frac{3 m_{\pi}}{2 p_0}\right ] ~~~~~~~(a3)\ ] ] performing now the average of the interference term over the isotropic distribution in @xmath454 : @xmath455 d \cos \theta_{\nu}^{\ast } \nonumber \\ & = & \frac{1}{2 } { \it re } \exp \left\{i \phi_{12}^{\nu,\pi}(0)+ \left(\frac{p_{\pi}}{m_{\pi}}\right)^2\left(i\frac{\delta m^2 l}{2 p_0 } \left[1-\frac{3 m_{\pi}}{2 p_0}\right ] \right .
. \nonumber \\ & & \left .
~~~~~~~~~~-\frac{1}{6}\left(\frac{\delta m^2 l}{p_0 } \left[1-\frac{m_{\pi}}{p_0}\right ] \right)^2\right ) \right\}~~~~~~~~~~~~~~~~~~(a4 ) \nonumber\end{aligned}\ ] ] in deriving eqn(a4 ) the following approximate formula is used : @xmath456 = \frac{\sin \alpha}{\alpha } \simeq 1-\frac{\alpha^2}{6}~~~~~~~~~~~~~~~~~~~~~~~~~(a5)\ ] ] where @xmath457 \ll 1\ ] ] the average over the maxwell - boltzmann distribution ( 2.31 ) is readily performed by ` completing the square ' in the exponential , with the result : @xmath458\right)^2 \right .
\nonumber \\ & & + \left .
i\left[\phi_{12}^{\nu,\pi}(0)+ \frac{3}{4}\left(\frac{\overline{p}_{\pi}}{m_{\pi}}\right)^2\left(\frac{\delta m^2 l}{p_0 } \left[\frac{3 m_{\pi}}{2 p_0}-1\right]\right)\right]\right\ } \nonumber \\ & \equiv & f^{\nu}(dop)\cos [ \phi_{12}^{\nu,\pi}(0)+\phi^{\nu}(dop ) ] \nonumber~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~(a6)\end{aligned}\ ] ] leading to eqns(2.32 ) and ( 2.33 ) for the doppler damping factor @xmath459 and phase shift @xmath460 , respectively .
the correction for the effect of thermal motion in the case of muon oscillations @xmath460 , respectively . is performed in a similar way .
the oscillation phase : @xmath461 is modified by the lorentz boost of the muon momentum and energy , and the relativistic time dilatation of the phase increment of the pion path , to : @xmath462 where @xmath463 and @xmath464 is given by eqn(2.37 ) . developing , as above , in terms of @xmath452 , gives : @xmath465 \cos\theta_{\mu}^{\ast } + \left(\frac{p_{\pi}}{m_{\pi}}\right)^2 \frac{m_{\mu}^2 \delta m^2 l}{p_0 ^ 3 } \left[\frac{e_0^{\mu}}{2 m_{\pi}}-1\right ] ( a9)\ ] ] performing the averages over @xmath466 and @xmath467 then leads to eqns(2.49 ) and ( 2.50 ) for the damping factor @xmath468 and phase shift @xmath469 , respectively .
the effect of the finite longitudinal dimensions of the target or detector is calculated by an appropriate weighting of the interference term according to the value of the distance @xmath470 between the decay and detection points ( see fig.1 ) .
writing the interference phase as @xmath471 , and assuming a uniform distribution of decay points within the target of thickness @xmath188 : @xmath472 substituting the value of @xmath0 appropriate to neutrino oscillations yields eqn(2.34 ) . since the value of @xmath0 is the same for neutrino and muon oscillations , the same formula is also valid in the latter case .
the same correction factor , with the replacement @xmath191 describes the effect of a finite detection region of length @xmath473 : @xmath474 the first step in the derivation of eqn(4.17 ) relating @xmath475 to @xmath160 is to calculate the angle @xmath476 , in the centre - of - mass ( cm ) system of the decaying pion , corresponding to @xmath276 in the laboratory ( lab ) system ( see fig.3 ) .
it is assumed , throughout , that the pion and muon are ultra - relativistic in the latter system , so that : @xmath477 .
the lorentz transformation relating the cm and lab systems gives the relation : @xmath478 the starred quantities refer to the pion cm system .
making the substitutions : @xmath479 , @xmath480 , eqns(b1 ) may be solved to obtain , up to first order in @xmath276 , @xmath476 and @xmath160 : @xmath481 where , ( c.f .
eqn(2.37 ) : @xmath482 using eqn(2.36 ) @xmath483 may be expressed in terms of the neutrino mass difference : @xmath484 eliminating now @xmath483 between ( b2 ) and ( b4 ) gives a relation between @xmath276 , @xmath476 and @xmath160 : @xmath485 in the lab system , and in the ur limit , the difference of the velocities of the muons recoiling against the two neutrino mass eigenstates is : @xmath486~~~~~~~~~~~~~~~~~(b6)\ ] ] where @xmath47 is the muon energy in the lab system for vanishing neutrino masses . making the lorentz transformation of the muon energy from the pion cm to the lab frames , and using eqns(2.4 ) and ( 2.36 ) to retain only terms linear in @xmath160 and @xmath476 , enables eqn(b6 ) to be re - written as : @xmath487 ~~~~~~~~~~~~~~~~~~~~~~~(b7)\ ] ] where @xmath45 is the energy of the pion beam . by combining the geometrical constraint equation for the muon velocities , ( 4.16 ) with ( b5 ) and ( b7 ) the angles @xmath276 and
@xmath476 may be eliminated to yield the equation for lab frame velocity difference : @xmath488 where @xmath489 @xmath490 to simplify ( b8 ) , the quantity @xmath491 is now expressed in terms of kinematic quantities in the pion cm system . within the ur approximation used , @xmath492 so that @xmath493 writing eqn(b1 ) to first order in @xmath264 , and neglecting terms of @xmath494 : @xmath495 using eqn(b12 ) , and expressing @xmath47 in terms of pion cm quantities , eqn(b11 ) may be written as : @xmath496 expressing the rhs of ( b13 ) in terms of @xmath45 and @xmath47 , using the relation : @xmath497 gives eqn(4.20 ) of the text . on substituting ( b13 ) into the rhs of ( b10 ) , it can be seen that the factor in the large curly brackets is the same in ( b9 ) and ( b10 ) , and so cancels in the ratio @xmath498 in eqn(b8 ) .
it follows that : @xmath499 finally , using ( b3 ) and ( b14 ) to express the factor in large brackets in eqn(b15 ) in terms of @xmath47 and @xmath45 , eqn(4.17 ) of the text is obtained . 99 b.pontecorvo , jetp * 33 * 599 ( 1957 ) , [ sov . phys .
jetp * 6 * 429 ( 1958 ) ] ; jetp * 34 * 247 ( 1958 ) , [ sov .
jetp * 7 * 172 ( 1958 ) ] .
s.m.bilenky and b.pontecorvo , phys . rep . * 41 * 225 ( 1978 ) . s.m.bilenky and s.t.petcov , rev .
phys . * 59 * 671 ( 1987 ) .
y.grossman and h.j.lipkin , phys . rev .
* d55 * 2760 ( 1997 ) . s. de leo , g.ducati and p.rotelli , _ comments upon the mass oscillation formulas _ , hep - ph/9906460 .
r.g.winter , lettere al nuovo cimento * 30 * 101 ( 1981 ) .
s.mohanty , _ covariant treatment of flavour oscillations _ , hep - ph/9702424 .
y.srivastava , a.widom and e.sassaroli , eur .
j * c2 * 769 ( 1998 ) .
r.p.feynman , phys . rev . *
76 * 749 ( 1949 ) .
et al . _ ,
phys . rev .
* d59 * 054018 ( 1999 ) .
a.d.dolgov _ et al . _ ,
b502 * 3 ( 1997 ) .
r.p.feynman , rev .
phys . * 20 * 367 ( 1948 ) .
p.a.m.dirac , physikalische zeitschrift der sowjetunion band 3 , heft 1 ( 1933 ) . reprinted in _
selected papers on quantum electrodynamics _
j.schwinger , dover , new york , ( 1958 ) p312 ; see also ref .
@xcite , chapter v , section 32 .
r.p.feynman and a.r.hibbs , _ quantum mechanics and path integrals _ , mcgraw hill , new york , ( 1965 ) ; c.grosche and f.steiner _ handbook of path integrals _ , springer tracts in modern physics , springer - verlag , berlin ( 1998 ) .
w.heisenberg , _ the physical principals of the quantum theory _ , english translation by c.eckart and f.c.hoyt , university of chicago press , chicago , 1930 .
chapter iv , section 2 .
p.a.m.dirac , _ the principles of quantum mechanics _ , fourth edition , o.u.p . , london ( 1958 ) p9 . c. athanassopoulos _ et al .
_ , phys . rev . *
c58 * 2489 ( 1998 ) ; a.aguilar _ et al .
_ , hep - ex/0104049 .
b.zeitnitz _ et al . _ ,
progress in particle and nuclear physics , * 40 * 169(1998 ) ; j.kleinfeller nucl .
b ( proc . suppl . ) * 87 * 281 ( 2000 ) .
l.i.schiff , _ quantum mechanics _ ,
2nd edition , mcgraw hill , new york , ( 1955 ) , chapter viii .
j.m.lvy-leblond and f.balibar , _ quantics , rudiments of quantum physics , _ north - holland , amsterdam , ( 1990 ) , chapter v. m.zralek , acta .
polon . , * b29 * 3925 ( 1998 ) .
c.giunti and c.w.kim , _ quantum mechanics of neutrino oscillations _ , hep - ph/0011074 .
c.giunti , c.w.kim and u.w.lee , phys . rev . *
d45 * 2414 ( 1992 ) .
v.gribov and b.pontecorvo , phys . lett . *
b28 * 493 ( 1969 ) . s.m.bilenky and b.pontecorvo , phys
b61 * 248 ( 1976 ) . h.fritsch and p.minkowski , phys
b62 * 72 ( 1976 ) . s.m.bilenky and b.pontecorvo , lettere al nuovo cimento * 17 * 569 ( 1976 ) . s.m.bilenky and b.pontecorvo , physics reports * 41 * 225 ( 1978 ) .
see ref .
@xcite , chapter ii .
b.kayser , phys . rev . *
d24 * 110 ( 1981 ) .
c.giunti , c.w.kim and u.w.lee , phys . rev . *
d44 * 3635 ( 1991 ) .
s.nussinov , phys . lett . *
b63 * 201 ( 1976 ) .
c.giunti , c.w.kim and u.w.lee , phys . lett . *
b274 * 87 ( 1992 ) .
k.kiers , s.nussinov and n.weiss , phys . rev . *
d53 * 537 ( 1996 ) .
m.beuthe , _ oscillations of neutrinos and mesons in quantum field theory _ , hep - ph/0109119 .
et al . _ ,
phys . rev .
* d48 * 4310 ( 1993 ) .
j.rich , phys . rev . *
d48 * 4318 ( 1993 ) . w.grimus and p.stockinger , phys
* d54 * 3414 ( 1996 ) .
j.e.campagne , phys . lett . *
b400 * 135 ( 1997 ) .
k.kiers and n.weiss , phys . rev .
* d57 * 3091 ( 1998 ) .
yu.v.shtanov , phys . rev . *
d57 * 4418 ( 1998 ) .
c.giunti and c.w.kim , phys . rev .
* d58 * 017301 ( 1998 ) .
c.giunti , c.w.kim and u.w.lee , phys . lett . *
b421 * 237 ( 1998 ) .
w.grimus , s.mohanty and p.stockinger , phys . rev . *
d59 * 013011 ( 1999 ) .
w.grimus , s.mohanty and p.stockinger , phys . rev . *
d61 * 033001 ( 1999 ) .
c.y.cardell , phys . rev . *
d61 * 073006 ( 1999 ) .
f.halzen and a.d.martin , _ quarks and leptons : an introductory course in particle physics _ , john wiley and sons , new york 1984 , section 4.8 .
m.blasone and g.vitello , ann .
( new york ) * 244 * 283 ( 1995 ) ; e * 249 * 363 ( 1995 ) .
m.blasone , p.a.henning and g.vitello , phys . lett . *
b451 * 140 ( 1999 ) . y.n.srivastava and a.widom , _ of course muons can oscillate _ , hep - ph/9707268 .
t.ohlsson , _ equivalence between neutrino oscillations and neutrino decoherence _ , hep - ph/0012272 .
b.kayser and l.stodolsky , phys . lett .
* b359 * 343 ( 1995 ) .
j.h.field , _ a covariant feynman path amplitude calculation of neutral kaon and b - meson oscillations _ , paper in preparation .
, d.e.groom _ et al .
j * c15 * , 1 ( 2000 ) . s.deleo and p.rotelli , _ timelapse _ , hep - ph/019014 . t.kajita and y.totsuka , rev .
phys . * 73 * 85 ( 2001 ) .
j.n.bahcall , m.h.pinsonneault and s.basu , astrophys . *
j 555 * 990 ( 2001 ) .
h.g.berry and j.l.subtil , phys .
* 27 * 1103 ( 1971 ) .
c.blondel , c.delsart and f.dulieu , phys .
lett . * 77 * 3755 ( 1996 ) .
_ , am j. phys .
* 66 * 38 ( 1998 ) .
c.blondel , s.berge and c.delsart , am j. phys . *
69 * 810 ( 2001 ) . | particle oscillations following neutrino production processes are analysed within feynman s path amplitude formulation of quantum mechanics .
consideration of the temporal sequence of production and detection events reveals an important contribution to the oscillation phase from the space - time propagator of the decaying source particle .
formulae are derived for spatial neutrino oscillations following pion decay ( at rest and in flight ) , muon decay and nuclear @xmath0-decay at rest , as well as for similar muon oscillations after pion decay at rest and in flight . in all cases
studied , the neutrino oscillation phases found differ from the conventionally used one .
a concise critical review is made of previous treatments of the quantum mechanics of neutrino and muon oscillations .
24.5 cm -5pt -5pt -50pt addtoresetequationsection * j.h.field * dpartement de physique nuclaire et corpusculaire universit de genve .
24 , quai ernest - ansermet ch-1211 genve 4 .
pacs 03.65.bz , 14.60.pq , 14.60.lm , 13.20.cz quantum mechanics , neutrino oscillations . |
two - dimensional gauge systems including generally covariant theories and those with higher rank fields strike a balance between descriptive simplicity and nontrivial content .
basically , this can be explained by that in two dimensions , unlike in higher dimensions , only scalar or spinor modes can propagate .
it follows that some of gauge systems are topological while others may exhibit a quite complicated dynamical structure which encodes matter modes .
both cases are intrinsically related because the matter can interact via topological gauge fields , while matter modes themselves can be represented as particular components of gauge fields .
this is most clearly illustrated by the massive schwinger model of fermions interacting via topological maxwell field , or by the liouville field originating as the metric component in the conformal gauge .
a prominent example of topological gauge systems is the jackiw - teitelboim gravity model @xcite which replaces the standard gravitation theory .
recall that in two dimensions einstein tensor vanishes identically so that the simplest alternative is given by diffeomorphism invariant equation of motion of the form is a general two - dimensional space - time with local coordinates @xmath3 , lorentz world indices run @xmath4 , lorentz fiber indices run @xmath5 , @xmath6 fiber indices run @xmath7 , @xmath6 invariant metric is @xmath8 .
the levi - civita tensor @xmath9 is normalized as @xmath10 .
two - dimensional anti - de sitter spacetime @xmath0 has a radius @xmath11 and a signature @xmath12 , so that the cosmological constant is @xmath13 .
the levi - civita tensor @xmath14 is normalized as @xmath15 . ]
[ jteq ] r-= 0 , where @xmath16 is the ricci scalar curvature built of the metric field @xmath17 , while @xmath18 is the cosmological constant .. ] the above theory is obviously non - lagrangian since one has a single equation for three variables .
adding an auxiliary scalar field @xmath19 allows one to introduce the variational principle so that the resulting local action describes a particular dilaton gravity .
the theory is however topological and contains no local degrees of freedom . at the same time , @xmath2 spacetime and black hole type objects are solutions of the jackiw - teitelboim model @xcite .
the main purpose of this paper is to elaborate a higher spin generalization of pure gravity system governed by equation .
we introduce higher rank gauge fields and their equations of motion in such a way that the resulting gauge system contains the gravitational subsector . as a guiding principle
, we employ the frame - like formulation instead of the metric - like one used to postulate equation .
the idea behind the frame - like approach originates from the cartan formulation of the einstein gravity where the metric is the symmetric component of the vielbein , while the antisymmetric component is a pure gauge by local lorentz transformations . indeed , when extending to higher spins , the jackiw - teitelboim model viewed as a gauge theory of the @xmath2 global symmetry algebra @xmath20 with a bf local action @xcite has conceptual advantages over alternative approaches .
one considers bf action with fields taking values in @xmath1 algebra with a finite parameter @xmath21 .
we show that such a theory can be interpreted as two - dimensional topological @xmath2 higher spin gravity theory .
note that conventional higher - spin metric - like fields @xmath22 with @xmath23 lorentz indices appear as particular tensor components of @xmath1 gauge connection rewritten in @xmath24 manifestly lorentz covariant basis .
other components are either auxiliary or pure gauges .
we find that the set of local fields over @xmath2 background consists of gauge fields with spins @xmath25 and masses @xmath26 interpreted as topological partially - massless fields of maximal depth .- form fields of the bf formulation will be analyzed elsewhere @xcite . ] on going from the original bf formulation linearized over @xmath2 space to the metric - like formulation one finds out that there are two different but dynamically equivalent forms of the metric - like theory .
in particular , it turns out that the gauge sector of the quadratic theory can be dually represented either by massive scalar equations or by conservation conditions for higher spin currents .
both scalar / current equations are invariant with respect to on - shell symmetries / improvements that guarantee the absence of local degrees of freedom . while the bf theories under consideration are topological , we have in mind to study @xmath2 models with local degrees of freedom interacting via higher spin topological fields .
such an interacting theory will contain at least cubic field - current interactions of the form @xmath27 , where @xmath28 are some lorentz spin-@xmath23 gauge fields , while @xmath29 are conserved lorentz spin-@xmath23 currents built of matter fields .
higher spin model has been proposed by vasiliev @xcite .
remarkably , it is formulated as bf theory with fields taking values in some infinite - dimensional algebra , and up to now this is the only example of fully non - linear higher spin gauge theory with local degrees of freedom formulated at the action level .
] we suggest that the field - current duality mentioned above will play an important role in the analysis of the higher spin field - current couplings in @xmath2 space .
it follows that one of the aims of the present paper is to classify possible higher spin gauge fields in @xmath2 space that may couple to conserved higher spin currents .
gravitational fields in two dimensions can be identified with @xmath6 connection @xmath30 , where @xmath31 are @xmath6 basis elements with the commutation relations @xmath32 = - \epsilon_{abc}\,t^c$ ] .. the commutation relations for antisymmetric basis elements @xmath33 take the standard form @xmath34 = \eta^{ad}t^{bc } + \text{3 terms } $ ] . ] adding @xmath35-form fields @xmath36 one arrives at the bf action @xcite [ jtaction ] s_jt[w , ] = _ ^2 _ a ^a , where the curvature is given by @xmath37 .
the equations of motion following from are [ zerc ] _
mn^a = 0 , d_m ^a = 0 , where @xmath38 is the @xmath6 covariant derivative .
the 1-form connection denoted as @xmath39 corresponds to @xmath0 spacetime .
the zero - curvature constraint @xmath40 expresses lorentz spin connection @xmath41 via the zweibein @xmath42 , while the latter yields @xmath0 spacetime metric @xmath43 through the standard identification @xmath44 , where the fiber minkowski metric is @xmath45 .
the bf frame - like equations of motion are dynamically equivalent to the metric - like equations of original jackiw - teitelboim model . in particular , the original equation is identified with one component of the curvature , while other two components are constraints ( see @xcite for more details ) . despite the general covariance , the theory describes a scalar liouville mode that follows from equation in the conformal gauge . however , the theory is topological because the resulting liouville equation possesses a residual coordinate invariance that gauges away all functional freedom of the solutions , see , _
e.g. , _ @xcite
. the higher spin generalization of the theory is straightforward .
two - dimensional bf theory with @xmath46-valued fields , where @xmath47 can be considered as particular higher spin gauge theory . in order to introduce higher spin gauge fields
one decomposes the adjoint of @xmath1 algebra into totally symmetric irreps of @xmath0 spacetime global symmetry algebra @xmath48 .
for instance , in the case of @xmath49 the basis elements @xmath50 labelled by @xmath51 can be equally rearranged as @xmath52 with @xmath53 , _
i.e. _ , as a direct sum of rank-@xmath54 and rank-@xmath55 totally symmetric and traceless @xmath56 algebra tensors .
in particular , respective one - form connections are identified with frame - like fields of spins @xmath55 and @xmath57 .
the analogous construction has been used in the context of @xmath58 chern - simons higher spin theory with algebra @xmath59 , where @xmath60 @xcite .
it follows that lie algebra @xmath1 for @xmath61 can be interpreted as a finite - dimensional higher spin algebra in two dimensions provided that its basis elements are organized as follows [ sln ] t_a_1 t_a_1a_2 t_a_1 ...
a_n-1 , where @xmath62 are rank-@xmath63 totally symmetric and traceless @xmath56 algebra tensors , [ irrgen ] t_(a_1 ... a_k ) : ^mn t_mna_3 ... a_k = 0 .
it is worth noting that the decomposition comes from the principal embedding of @xmath64 .
chern - simons higher spin theory non - principal embeddings and respective spectra were discussed , _
e.g. , _ in @xcite . ]
introduce differential 0-form and 1-form fields taking values in @xmath65 with basis elements [ connec ] ( x ) = _
a_s-1(x ) t_a_1 ... a_s-1 , w(x ) = _
s=2^n-1dx^m w_m^a_1 ... a_s-1(x ) t_a_1 ... a_s-1 .
expansion coefficients in the higher spin algebra basis elements are to be identified as higher spin fields .
they satisfy @xmath6 irreducibility conditions of the form .
let us define @xmath2 higher spin gravity by the following bf action [ action ] s [ w , ] = g _ s=2^n-1 _ ^2 _ a_1 ...
a_s-1 ^a_1 ...
a_s-1 , with curvature two - forms given by @xmath66 } + [ w_m , w_n]^{a_1 ... a_{s-1}}$ ] , where @xmath67 $ ] is the lie bracket in @xmath46 evaluated in basis , and @xmath68 is a dimensionless coupling constant . the action is invariant under the @xmath69 gauge transformations [ fulgaugetr ] w_m^a_1 ... a_s-1= _ m ^a_1 ... a_s-1 + [ , w_m]^a_1 ... a_s-1 , ^a_1 ...
a_s-1 = [ , ] ^a_1 ... a_s-1 , where @xmath70 is an @xmath47-valued gauge parameter .
note that _ a - priori _ there is no metric tensor as the theory is completely formulated via differential forms .
it follows that the theory is diffeomorphism invariant .
however , an on - shell diffeomorphism transformation is simply a gauge transformation with particular gauge parameter .
variational equations of motion for @xmath71 and @xmath72 are the zero - curvature equation and the covariance constancy equation [ curveq ] _ mn^a_1 ...
a_s-1= 0 , d_m ^a_1 ... a_s-1 = 0 , s = 2 , ... , n-1 , where @xmath73 is a covariant derivative with 1-form connection @xmath74 given by .
the covariance constancy equation involves both 0-form and @xmath54-form fields , while the zero - curvature equation forms an independent gauge sector of the theory that can be analyzed separately .
let us emphasize that solutions of the zero - curvature equation are given locally by pure gauge 1-form fields . for the case of
finitely many gauge fields this observation guarantees that the theory is topological and therefore contains no local degrees of freedom .
however , in the next section we explicitly examine the gauge sector in the free field approximation to provide a particular physical interpretation of topological modes .
the higher spin theory contains a gravitational sector described by the jackiw - teitelboim action .
the gravitational fields are singled out by setting all higher spin fields to zero except for those taking values in @xmath75 subalgebra .
this particular truncation is obviously consistent . at the same time , @xmath2 space is the ground state of the theory since it solves gravitational equations .
since the action is non - linear containing at most cubic terms @xmath76 , one can consider the perturbation theory over a particular solution . to this end
introduce 1-form @xmath77 describing @xmath0 spacetime and 0-form @xmath78 as the background of the theory , so that all fields are decomposed as @xmath79 and @xmath80 , where @xmath81 and @xmath82 are fluctuations over the background .
the background fields @xmath77 and @xmath83 obviously solve bf higher spin equations . similarly , the 2-form curvatures are expanded as @xmath84 , where @xmath85 is the zero - curvature gravitational constraint , while @xmath16 is the linearized curvature and the ellipsis denote second order corrections . in the one - form sector ,
gauge fields @xmath81 and their linearized curvatures @xmath16 have the component form [ gaugefield ] ^a_1 ...
a_s-1 = dx^m _ m^a_1 ... a_s-1 , r^a_1 ...
a_s-1 = d_0^a_1 ... a_s-1 , where @xmath86 is the background covariant derivative . representing the zero - curvature condition as @xmath87 one obtains that linearized curvatures are invariant under the gauge transformations expanded over the background , [ trans ] ^a_1 ...
a_s-1 = d_0 ^a_1 ...
a_s-1 , where @xmath88 are 0-form gauge parameters . the bianchi identities @xmath89 following from are trivial since any 3-form in two dimensions vanishes identically .
fluctuations in the zero - form sector are inert under the gauge transformations @xmath90 , at least to the lowest order in the fields , cf . .
let us compare the resulting set of 1-form higher spin gauge fields in @xmath91 dimensions with fields in @xmath92 dimensions .
by massless higher spin fields in @xmath93 we call 1-form connections @xmath94 taking values in @xmath95 finite - dimensional irreps of the symmetry type described by two - row rectangular young diagram @xcite . in the metric - like formulation
these fields reduce to fronsdal spin-@xmath23 massless fields .
obviously , for @xmath96 all 1-form connections vanish identically except for @xmath97 case of the gravity , @xmath98 .
it follows that there are no spin @xmath99 fronsdal fields in @xmath2 . at the same time , in two dimensions fiber indices of spin-@xmath55 connection can be dualized by virtue of @xmath6 levi - civita tensor as @xmath100 .
the point is that the gravitational connection @xmath101 starts an infinite sequence of 1-form connections @xmath102 , @xmath103 , taking values in @xmath6 finite - dimensional irreps described by one - row young diagrams .
these are fields appearing via the gauging of @xmath1 algebra in the basis .
note that in the @xmath104 case , the connections with totally symmetric @xmath105 fiber indices describe partially - massless fields of maximal depth @xcite .
it follows that in @xmath2 space all possible higher spin @xmath99 fields are exhausted by `` topological partially - massless '' fields of maximal depth . dynamical field , @xmath106 . in two dimensions ,
a maximal value of the depth equals @xmath107 due to the hodge duality mentioned above . ]
the bf equations linearized around @xmath2 background space decouple into a direct sum of @xmath108 independent subsectors [ 1eq ] d_0 ^a_1 ... a_s-1 = 0 , [ 2eq ] d_0 ^a_1 ... a_s-1 = 0 , each of spin @xmath109 . in
what follows , we analyze a set of topological metric - like fields encoded in the gauge sector described by equations . to identify the dynamical content of equations
one decomposes @xmath6 covariant gauge fields and linearized curvatures into lorentz algebra @xmath110 irreducible components , [ dectraceless ] _
a_s-1 = _ k=0^s-1 _ m^a_1 ... a_k , r_mn^a_1 ... a_s-1 = _
k=0^s-1 r_mn^a_1 ... a_k , _ i.e. _ , totally symmetric and traceless with respect to the minkowski tensor @xmath111 . assuming that lorentz indices @xmath112 are symmetrized with a unit weight , we find that the component form of is given by [ curva ] r_mn^a_1
... a_k = ^_[m _ n]^a_1 ...
a_k - h^_[m_,c _
n]^ca_1 ... a_k + + + + , where covariant derivative @xmath113 is defined with respect to the background lorentz spin connection @xmath41 , while @xmath114 is the background zweibein .
quite analogously , one finds the component form of the gauge transformations , where gauge parameters @xmath115 are @xmath116 irreducible components of the @xmath6 gauge parameter @xmath117 .
the lorentz component form of bf equations is [ main ] r^a_1 ... a_k_mn = 0 , k = 0 , 1 , ... , s-1 . in the spin @xmath118 case the only field @xmath119 is identified with potential @xmath120 .
the resulting bf equation @xmath121 is different form the maxwell equation @xmath122 .
it is worth noting here that a similar difference is between maxwell and chern - simons equations in three dimensions .
the analysis of equations for spin @xmath123 is more intricate because there are more different rank fields and their gauge transformations including both differential and algebraic symmetry parameters .
all fields entering equations of motion can be divided into three subsets consisting of dynamical fields , stueckelberg fields , and auxiliary fields .
it turns out that such a triple decomposition of the original field space can be done in two different ways . in the following sections we explicitly demonstrate how it works in the spin-@xmath55 case , and
then briefly generalize to arbitrary spins . in the spin @xmath97 case fields @xmath124 and
@xmath119 are identified with the zweibein and lorentz spin connection ( more precisely , with their fluctuations over @xmath2 values @xmath125 and @xmath41 ) . field equations read off from and at @xmath97 [ r ] r_mn _ m _ n - _ n _ m - h_m_,b_n^b + h_n_,b _ m^b = 0 , and [ ra ] r_mn^a _ m ^a_n - _
n ^a_m + h_m^a _ n - h_n^a _ m = 0 , are invariant under the gauge transformations [ gt ] _ m = _ m -h_m_,c ^c , _
m^a = _ m ^a + h_m^a . using dualized field redefinitions @xmath126 and @xmath127}$ ] , the equations and can be cast into the standard form for zweibein @xmath128 and lorentz spin connection @xmath129 .
in particular , the above field redefinition succeeded by setting @xmath130 reproduces the known expressions for curvatures of the @xmath131 poincare algebra .
the standard choice of dynamical fields is achieved by using the gauge parameter @xmath132 as the stueckelberg one to gauge away the trace component of @xmath133 .
can be dualized into an antisymmetric parameter @xmath134}$ ] .
obviously , this choice corresponds to the standard gauge fixing of local lorentz symmetry . ] to this end , using the @xmath2 background frame @xmath42 one converts world indices into fiber ones and then decomposes @xmath133 into three @xmath116 irreducible components as follows [ decomp ] ^a|b = ^(ab ) + ^ab + ^ab , where @xmath135 .
residual components of @xmath124 can be combined into a single traceful symmetric tensor @xmath136 , so that its traceless and trace parts are given by @xmath137 and @xmath138 .
field @xmath139 is then identified as dynamical field with the gauge symmetry transformation @xmath140 ( cf .
our comments below ) .
simultaneously , one considers equation as the constraint which allows one to express auxiliary field @xmath119 via dynamical field @xmath139 .
equation @xmath141 is dynamical . in terms of @xmath142
it reads @xmath143 , where @xmath144 is the wave operator for the @xmath2 metric , and @xmath145 .
obviously , this is the jackiw - teitelboim equation linearized around @xmath2 background . imposing the conformal gauge which in our case says that the traceless part of @xmath139 is set to zero , one is left with @xmath146 subjected to the residual equation [ liouvilleliner ] ( _ ads_2 - 2)= 0 .
the above equation is invariant under the residual gauge transformation @xmath147 with a parameter satisfying the killing equation @xmath148 .
therefore , one is left with the scalar field which is massive gauge field .
it is crucial that for @xmath149 the above choice of dynamical fields is not unique .
another option is to use the gauge parameter @xmath150 as a stueckelberg one to completely gauge away the field @xmath151 .
then , from and one derives the algebraic constraint @xmath152 , which says that the antisymmetric component of @xmath133 is set to zero .
it follows that a dynamical field is again a traceful symmetric tensor @xmath153 , but now its traceless and trace components are identified with @xmath137 and @xmath154 .
the gauge transformations now read @xmath155 .
note that it reproduces the gauge transformation of a spin-@xmath55 partially - massless field .
equation @xmath156 is dynamical . imposing the gauge @xmath157 one
is left with a traceless component subjected to the residual equation [ conser ] ^a j_ab = 0 , where @xmath158 is symmetric and traceless tensor field . a residual gauge invariance of is given by @xmath159 with the scalar parameter satisfying differential constraint @xmath160 .
it follows that equation can be seen as the conservation condition for a rank-@xmath55 current @xmath161 on the @xmath2 space . by the standard realization we mean using a set of dynamical fields that generalize the metric - like variable of the spin-@xmath55 case
a careful analysis shows that all higher rank equations in are constraints except for the @xmath162 equation @xmath141 treated as the dynamical equation .
the constraints allow one to express all fields via derivatives of dynamical fields which are scalar and rank-@xmath23 traceless tensors , @xmath146 and @xmath163 . as in the spin-@xmath55 case , the scalar component @xmath146 is identified with antisymmetric component of @xmath124 .
the rank-@xmath23 traceless component @xmath164 is identified with a totally symmetrized highest rank @xmath165 connection @xmath166 .
note that the symmetrized combination is automatically traceless because all other components in @xmath167 are gauged away using the stueckelberg symmetry transformations .
quite analogously , one shows that the only independent gauge parameter is @xmath168 , while all others are treated as stueckelberg parameters or auxiliary parameters expressed via derivatives of @xmath168 .
the leftover gauge symmetry with parameter @xmath168 allows one to impose the higher spin gauge , [ hsgc ] ^a_1 ... a_s = 0 , which is legitimate since any @xmath116 rank-@xmath63 irreducible tensor for @xmath169 has just two independent components .
then , the only dynamical field is given by the scalar @xmath146 . imposing the higher spin gauge and solving the constraints in one finds that the leftover equation reduces to the massive scalar equation with a particular value of mass - like term [ scaleq ] ( _ ads_2- m_s^2)= 0 , m_s^2 = s(s-1),s2 .
this is invariant under the residual gauge transformations , [ leftgauge1 ] = _ mn_1^m _ n_2 _ n_s-1^n_1 n_s-1 , provided that the gauge parameter satisfies the generalized killing equation @xmath170 , which is simply the stability transformation for the higher spin gauge .
the massive scalar field endowed with higher order derivative transformations for the scalar parameter can be treated as topological partially - massless field of spin @xmath23 and the maximal depth . by the dual realization we mean using a set of dynamical fields that generalize the current - like variable of the spin-@xmath55 case .
quite analogously to the standard realization one shows that all higher rank equations in can be treated as constraints except for the @xmath165 equation @xmath171 which is dynamical .
the constraints allow one to express all fields via derivatives of dynamical fields also given by scalar and rank-@xmath23 symmetric traceless tensor , @xmath172 and @xmath173 .
but now the scalar @xmath146 is identified with a trace component of @xmath124 , while the rank-@xmath23 traceless field @xmath173 is identified with the symmetric traceless component of @xmath165 connection @xmath174 , namely @xmath175 .
an independent gauge parameter is the scalar @xmath132 , while the gauge transformations read [ residualtr ] = _ ads_2-s(s-1 ) , _
a_1 ... a_s = _
a_s + ... , where the ellipsis refers to proper symmetrizations and trace terms .
using gauge symmetry one can impose the scalar gauge @xmath176 along with the residual gauge parameter equation @xmath177 . in the scalar gauge ,
the only dynamical field is the rank-@xmath23 component @xmath178 .
then , solving the constraints in one finds that the independent dynamical equation is given by [ inteq ] ^n j_n a_1 ... a_s-1 = 0 j_na_1 ...
n^m_ma_1 ... a_s-1 , where dualized tensor field @xmath179 is totally symmetric and traceless .
one identifies @xmath180 with spin-@xmath23 conserved current on the @xmath2 space , while higher order derivative transformations in are `` improvements '' . on the other hand ,
the `` improvement '' transformation can treated as the gauge transformation of a spin-@xmath23 partially massless field of maximal depth .
in this paper we have introduced @xmath2 higher spin gravity defined via topological bf action with fields taking values in @xmath1 algebra .
we identified the spectrum of higher spin gauge field fluctuations over @xmath2 background , and for @xmath149 found out their dual description in terms of higher rank conserved currents .
note that the proposed duality between two metric - like formulations guarantees that the two realizations describe the same dynamical system . indeed , in order to single out independent fields and parameters we used stueckelberg algebraic symmetry and eliminated auxiliary fields of the frame - like formulation .
whence , two metric - like formulations and coming from the original ( linearized ) bf system are dynamically equivalent .
this type of duality is similar to that one in the wznw theory with the equations of motions for a group variable @xmath181 given by a second order field equation @xmath182 . at the same time , introducing the current @xmath183 one equivalently arrives at the first order conservation condition @xmath184 ( _ e.g. _ , see @xcite ) .
it is instructive to compare linearizations of @xmath2 higher spin theory for @xmath47 algebra and @xmath58 chern - simons higher spin theory for @xmath185 algebra . in the @xmath58 case , choosing @xmath186 spacetime as the background for higher spin fields yields fronsdal equations of motion for massless gauge fields with spins @xmath187 that describe , however , no local degrees of freedom @xcite . in the @xmath188 case
, the same analysis does not yield fronsdal equations , although the system is also topological . instead , in the higher spin gauge which generalizes the conformal gauge in the gravity , the dynamical equations in the standard realization are scalar field equations with higher order derivative gauge symmetries guaranteeing the absence of propagating modes .
these gauge systems can be interpreted as two - dimensional topological partially - massless fields .
it is remarkable that bf higher spin theory considered in the present paper has been in fact formulated within the unfolded approach to higher spin dynamics . in @xcite
we consider the theory from the unfolded formulation perspective and use the cohomological technique to analyze bf equations of motion .
in particular , independent parameters , fields and equations of the standard and the dual realizations elaborated in sections * [ sec : sigmaplus ] * and * [ sec : sigmaminus ] * have a streamlined interpretation as cohomology elements of the so - called @xmath189 and @xmath190 nilpotent operators acting in the field space . also , using the cohomological machinery we analyze equations for 0-forms which were not considered in the present paper .
the proposed bf higher spin theory can be extended so as to include an infinite tower of higher spin fields provided a gauge algebra @xmath46 is chosen to be one of infinite - dimensional higher spin algebras @xcite .
the resulting theory is also topological .
recall that bf higher spin theory can however have local degrees of freedom provided that fields take values in particular infinite - dimensional associative algebra @xcite .
* acknowledgements .
* i am grateful to m. grigoriev , e. skvortsov and m.a .
vasiliev for useful comments and discussions .
i would like to thank the galileo galilei institute for theoretical physics , florence , italy , for hospitality during my staying at ggi workshop on higher spins , strings and duality , and to the dynasty foundation for the financial support of the visit .
the work is supported by rfbr grant 12 - 02 - 31838 . * note added . *
after the present paper has been submitted to arxiv , we were informed by soo - jong rey that @xmath188 higher spin theories were partly discussed in his talk @xcite , mainly from the holographic perspective . m. cadoni and s. mignemi ,
`` nonsingular four - dimensional black holes and the jackiw - teitelboim theory , '' _
_ * d51 * ( 1995 ) 43194329 , http://www.arxiv.org/abs/hep-th/9410041[hep-th/9410041 ] .
m. spradlin and a. strominger , `` vacuum states for ads(2 ) black holes , '' jhep * 9911 * ( 1999 ) 021 http://www.arxiv.org/abs/hep-th/9904143[hep-th/9904143 ] .
t. fukuyama and k. kamimura , `` gauge theory of two - dimensional gravity , '' _
_ * b160 * ( 1985 ) 259 .
k. isler and c. trugenberger , `` a gauge theory of two - dimensional quantum gravity , '' _ phys.rev.lett . _
* 63 * ( 1989 ) 834 .
a. h. chamseddine and d. wyler , `` topological gravity in ( 1 + 1)-dimensions , '' _ nucl .
_ * b340 * ( 1990 ) 595616 .
m. a. vasiliev , `` higher - spin gauge interactions for matter fields in two dimensions , '' _ phys .
* b363 * ( 1995 ) 5157 , http://www.arxiv.org/abs/hep-th/9511063[hep-th/9511063 ] .
r. jackiw , `` gauge theories for gravity on a line , '' _ theor.math.phys . _
* 92 * ( 1992 ) 979987 , http://www.arxiv.org/abs/hep-th/9206093[hep-th/9206093 ] .
m. henneaux and s .- j .
rey , `` nonlinear @xmath191 as asymptotic symmetry of three - dimensional higher spin anti - de sitter gravity , '' _ jhep _ * 1012 * ( 2010 ) 007 , http://www.arxiv.org/abs/1008.4579[1008.4579 ] .
a. campoleoni , s. fredenhagen , s. pfenninger , and s. theisen , `` asymptotic symmetries of three - dimensional gravity coupled to higher - spin fields , '' _ jhep _ * 1011 * ( 2010 ) 007 , http://www.arxiv.org/abs/1008.4744[1008.4744 ] .
a. castro , e. hijano , and a. lepage - jutier , `` unitarity bounds in ads(3 ) higher spin gravity , '' _ jhep _ * 1206 * ( 2012 ) 001 , http://www.arxiv.org/abs/1202.4467[1202.4467 ] .
m. a. vasiliev , `` cubic interactions of bosonic higher spin gauge fields in ads(5 ) , '' _ nucl .
* b616 * ( 2001 ) 106162 , http://www.arxiv.org/abs/hep-th/0106200[hep-th/0106200 ] .
s. deser and a. waldron , `` partial masslessness of higher spins in ( a)ds , '' _
_ * b607 * ( 2001 ) 577604 , http://www.arxiv.org/abs/hep-th/0103198[hep-th/0103198 ] .
zinoviev , `` on massive high spin particles in ads , '' http://www.arxiv.org/abs/hep-th/0108192[hep-th/0108192 ] .
c. r. nappi , `` some properties of an analog of the nonlinear @xmath192 model , '' _ phys.rev . _ * d21 * ( 1980 ) 418 ; e. fradkin and a. a. tseytlin , `` quantum equivalence of dual field theories , '' _ annals phys . _ * 162 * ( 1985 ) 31 .
k. alkalaev , `` global and local properties of @xmath0 higher spin gravity , '' preprint fian
td201316 , to be published .
e. fradkin and v. y. linetsky , `` higher spin symmetry in one - dimension and two - dimensions , i , '' _ mod.phys.lett .
_ * a4 * ( 1989 ) 26352647 .
soo - jong rey , `` news from higher spins : @xmath193 , black holes and entropy '' , talk given at the workshop `` higher spins and holography '' , simons center , march 2011 . | we formulate @xmath0 higher spin gravity as bf theory with fields taking values in @xmath1 algebra treated as higher spin algebra .
the theory is topological and naturally extends the jackiw - teitelboim gravity model so as to include higher spin fields .
the bf equations linearized about @xmath2 background are interpreted as describing higher spin partially - massless fields of maximal depth along with dilaton fields .
it is shown that there are dual metric - like formulations following from the original linearized bf higher spin theory .
the duality establishes a dynamical equivalence of the metric - like field equations that can be given either as massive scalar field equations or as conservation conditions for higher spin currents .
fian - td-2013 - 10 + * on higher spin extension of + the jackiw - teitelboim gravity model * k.b .
alkalaev _ i.e. tamm department of theoretical physics , + p.n .
lebedev physical institute , + leninsky ave .
53 , 119991 moscow , russia _ |
spin - orbit interactions which are present in quasi one - dimensional ( 1d ) semiconductor heterostructures like @xmath7 quantum wires provide a promising way to controllable spin manipulations@xcite . besides the rashba spin - orbit interaction @xmath8 @xcite , which is induced by an electric field , one deals with the linearized dresselhaus spin - orbit interaction @xmath9 @xcite .
these interactions rely on the presence of a structural and crystal inversion symmetry , respectively .
superpositions of such interactions under the influence of transversal@xcite and in - plane magnetic fields@xcite like @xmath10 , have also been discussed . the polar angle between the magnetic field and the @xmath11-axis is denoted by @xmath12 .
the corresponding couplings are denoted by @xmath13 and @xmath14 , in which case the equal coupling strength regime@xcite proceeds via @xmath15 . a typical choice is given by @xmath16 , or equivalently , @xmath17 , where @xmath18 denotes the dimensionless spin - orbit coupling .
equal coupling strength superpositions of such interactions in the presence of in - plane magnetic fields have also been analyzed recently@xcite . in this later case
it has been shown that accounting for spin conservations amounts to the selection of the polar angle along the bisectrices .
having obtained the energy band opens the way to the theoretical derivation of novel spin precession effects , as shown by ( 72 ) in ref .
9 . for this purpose one
proceeds along the lines presented before@xcite , but now by resorting to a different energy band structure .
this amounts to reverse the usual @xmath2-wavenumber dependence of the energy in order to establish two correlated wavenumbers , say @xmath19 and @xmath20 , which are responsible for the description of propagation paths along the @xmath11-axis .
then the two dimensional rotation matrix@xcite one looks for can be established in terms of displacements of length @xmath6 acting along the two paths just referred to above@xcite .
however , several details referring to a systematic study of the parameter dependent spin precession angles are still desirable .
the suitable selection of the @xmath2-parameter deserves a little bit more attention , too .
besides the influence of the discrete parameter @xmath21 , which is responsible for the @xmath22-signs in the front of the square root of the energy , we shall account this time for a further parameter , say @xmath23 , reflecting the rescaling of squared wavenumber via @xmath24 .
handling the parameter dependence of spin precession angles established in this manner amounts to deal with interplays between @xmath2 , @xmath1 , @xmath0 , @xmath25 and @xmath23 , which represents our main motivation in this paper .
so far numerical @xmath2-inputs are introduced via @xmath5 , where @xmath6 denotes specifically the nanometer length scale of the quantum wire model .
forbidden regions can be readily established in terms of selected parameters for which the precession angles are imaginary .
this happens in configurations for which @xmath26 .
complementary intervals should then be responsible for admissible configurations .
the @xmath23-parameter opens the way to extrapolations of the wavenumber towards imaginary and complex values , which looks promising for further generalizations .
the paper is organized as follows .
preliminaries and notations are discussed in sec .
the present hamiltonian is introduced by neglecting , for convenience , the orbital effects of the magnetic field .
accounting for spin conservations leads to the selection of @xmath27two @xmath12-angles , namely of @xmath28 and @xmath29 in which case @xmath30 and @xmath31 , respectively .
then the equal strength limit of the energy can be readily established .
this leads in turn to the displacement momenta @xmath32 @xmath33 serving to the description of two paths one looks for , as indicated in sec .
iii . in sec .
iv one shows that such momenta work safely whenever @xmath34 , but suitable convergence conditions have to be accounted for in so far as @xmath26 .
precession angles established before to first @xmath35-order via @xmath36 are discussed in sec . v. numerical studies concerning these angles are presented in some detail in sec .
vi . section vii deals with imaginary and complex realizations of the @xmath2-parameter .
the conclusions are presented in sec .
viii . mathematical details concerning the derivation of precession angles are shortly reviewed in the appendix .
in order to perform the quantum theoretical description of electronic behavior in two - dimensional ( 2d ) semiconductor heterostructures , single particle hamiltonians@xcite @xmath37 including the spin - orbit interactions have been proposed .
this time the orbital effects of the magnetic field have been neglected .
the momentum operator reads @xmath38 , whereas @xmath39 stands for the effective mass of the electron .
one has e.g. @xmath40 for @xmath7 quantum wires@xcite , where @xmath41 denotes the usual rest - mass of the electron .
the wire geometry is characterized by a transversal confining potential , say @xmath42 .
the zeeman interaction @xmath43 has also been incorporated , where @xmath44 stands for the bohr - magneton , while @xmath45 denotes the effective gyromagnetic factor . for convenience
, we shall assume that @xmath46 , as usual . using the total hamiltonian displayed in ( 1 ) leads to commutation relations like @xmath47 = \frac{2i\sigma _ { z}}{\hbar } \left ( p_{x}\pm p_{y}\right ) \left ( \alpha _ { r}\pm \alpha _ { d}\right ) -i\sigma _ { z}g\mu _ { b}b\left ( \sin \theta \mp \cos \theta \right ) \quad\ ] ] which exhibit the zero value
if @xmath48 so , one gets faced with conserved spin observables like @xmath49 and @xmath50 when the rashba and dresselhaus couplings exhibit the same magnitude , i.e. @xmath31 and @xmath51 respectively .
this proceeds in conjunction with selected in - plane orientations of the magnetic field @xmath52 for which @xmath53 and @xmath54 , respectively , as displayed above .
the energy eigenvalue problem can be solved by resorting once more again to the zero determinant condition for a homogenous system of two coupled equations , as shown many times before@xcite .
starting from the wavefunction @xmath55 where @xmath56 stands for the oscillator eigenfunction , yields the 1d - reduction of ( 1 ) as @xmath57 in which the oscillator eigenvalue reads @xmath58 , as usual ( see e.g.ref . [ 13 ] ) .
now we are ready to apply the zero determinant condition referred to above , which produces the energy @xmath59 in accord with ( 16 ) in@xcite , where @xmath60 and where the @xmath22-signs in the front of @xmath61 can also be viewed , in general , as reflecting the influence of the spin@xcite .
we then have to introduce a further parameter like @xmath21 , which will be used hereafter .
accordingly , the lower energy configuration corresponds to @xmath26 .
the corresponding spinorial eigenfunction is given by @xmath62 and @xmath63 , where @xmath64 the equal strength limit of ( 8) can also be readily performed .
one would then obtain @xmath65 , in accord with ( 3 ) , so that @xmath66 .
this means that @xmath67 @xmath68 if @xmath69 @xmath70 .
accounting for the spin conservation , we have to realize that the symmetrized equal coupling strength limit of the present spin dependent but non - symmetrized energy band ( 6 ) is given by @xmath71 by virtue of ( 3 ) .
so far , the dimensionless spin - orbit coupling , i.e. @xmath72 , has the magnitude order of unity .
one sees that the linear @xmath1-dependent term under the square root in ( 6 ) is ruled out , which comes definitely from the inter - related equal coupling strength limit one deals with in this paper .
moreover , ruling out the @xmath1-dependent term leads to the conversion of ( 6 ) into a conditionally solvable biquadratic equation in @xmath73 , which serves as a starting point to the derivation of spin precession effects .
rescaled variables like @xmath74 , @xmath75 , @xmath76 and @xmath77 can also be introduced .
numerical studies can then be readily done by starting from @xmath78 , @xmath79 , @xmath80 , @xmath81 and @xmath82 .
we shall also assume that @xmath83 @xcite .
we have to keep in mind that present calculations are sensitive to the numerical selection of @xmath2-parameter .
we have to realize that such selections can be established reasonably via@xmath84 , which yields a typical nanometer length like @xmath85 , when @xmath86 .
however , we have to be aware that other selections , @xmath2-dependent ones included , are conceivable . for convenience
, we shall insert hereafter the wavenumber input @xmath87 , though the choice @xmath88 has been used tentatively before@xcite .
it is also clear that ( 9 ) can be rewritten equivalently as @xmath89 ^{1/2}\ ] ] where @xmath21 , which produces the factorized algebraic equation @xmath90 where @xmath91 stand implicitly for the roots serving to the description of two propagation paths .
next we shall resort to a further approximation , namely to handle in the sequel the @xmath92-functions in terms of numerical inputs for the @xmath2-parameter .
then we are in a position to establish actually @xmath2-roots one looks for as @xmath93 where @xmath94 ^{1/2}}{\omega _ { 0}^{2}}% > 0\ ] ] and @xmath95 the same job can be done by resorting to numerical energy - inputs , such as indicated by the equation @xmath96 accordingly , one obtains @xmath87 when @xmath97 .
one remarks that ( 12 ) produces a power series in terms of the convergence condition @xmath98 , which is synonymous to @xmath99 this inequality is fulfilled automatically if @xmath34 .
however one gets faced with extra conditions like @xmath100 if @xmath26 , provided that @xmath101 next , we shall handle ( 17 ) via @xmath102 , @xmath103 and @xmath104 , where @xmath105 stands for a small parameter .
such conditions are reminiscent to the asymptotic description of nonlinear oscillations@xcite .
the interesting point is that ( 17 ) provides an admissible but finite @xmath1-interval such as given by @xmath106 in so far as the @xmath73-term is neglected , where @xmath107 it should be noted that @xmath0 plays this time the role of an input parameter .
conversely , starting from an input @xmath1-parameter leads to two disjoint but admissible semi - infinite @xmath0-intervals like @xmath108 and @xmath109 so that ( 18 ) gets converted into @xmath110 concrete realizations concerning ( 19 ) and ( 23 ) will be presented below . under such conditions
the leading approximations characterizing ( 12 ) are given by @xmath111 both for @xmath34 and @xmath26 , where @xmath112 , with the understanding that in the latter case the wavenumber description proceeds in terms of ( 19 ) and ( 23 ) .
displacements of length @xmath6 along the @xmath11-axis can be readily applied by resorting to the orthonormalized spinor @xmath113 by virtue of ( 4 ) .
this results in a 2d rotation matrix [ 10,11 ] , providing in turn the precession angle as @xmath114 ^{1/2}\ ] ] as shown in the appendix , which proceeds in accord with ( 72 ) in ref .
our main emphasis in this paper is one refs .
9 and 10 , but we have to mention that starting with the idea of the voltage controlled spin precession@xcite , a multitude of spin precession descriptions have also been presented during time@xcite , numerical studies included@xcite .
now we are ready to discuss the parameter dependence of the precession angle in a more systematic manner . inserting @xmath87 ,
provides in turn the @xmath115- and @xmath0-dependence of the precession angle @xmath116 in terms of @xmath117- and @xmath1-inputs respectively .
one finds that @xmath116 is a positive concave function of @xmath1 and @xmath0 increasing monotonically with @xmath118 and @xmath119 , respectively , in so far as @xmath34 .
however , one gets faced with nontrivial patterns when @xmath26 .
so there are crossing points between @xmath120-plots for @xmath34 and @xmath26 concerning both @xmath1- and @xmath0-dependent curves .
such points are located at @xmath121 and @xmath122 respectively .
these points reproduce identically the edge points characterizing the admissible intervals ( 19 ) and ( 23 ) .
it should be noted that both ( 27 ) and ( 28 ) are produced by the basic equation @xmath123 which has the meaning of a leading approximation .
the upper bounds characterizing the precession angle within the admissible intervals are then given by @xmath124 and @xmath125 where @xmath0 and @xmath1 stand for inputs .
other characteristic point of interest are the zeros exhibited by the @xmath26-precession angle on the @xmath1- and @xmath0-axes .
one realizes that such zeros are inter - related with the onset of discontinuity points of the second kind such that @xmath126 , as displayed in figs . 1 - 4 .
handling this latter equation gives @xmath127 which yields the solutions @xmath128 or @xmath129 ^{1/2}\right ] ^{1/2}\in \widetilde{% i}_{a}(b)\ ] ] respectively .
it can be easily verified that the zeros established in this manner get included into the admissible intervals ( 19 ) and ( 23 ) .
we have to remark that the that the zeros of the denominator in ( 31 ) such as given by @xmath130 produce vertical singularity lines which are asymptotically indistinguishable from supercritical @xmath120-tails for which @xmath131 and @xmath132 , which proceed in connection with @xmath133 and @xmath134 , respectively .
the understanding is that ( 29 ) is a leading approximation which comes from ( 35 ) via @xmath135 .
such vanishingly small strips tails lying outside admissible regions should then be viewed as meaningless artifacts which can be hereafter ignored . under such conditions
one gets faced with a leading approximation for which @xmath136 , which opens the way to a reasonable synthesis of admissible @xmath120-trajectories in pertinent parameter spaces .
after having been arrived at this stage , we are in a position to say that forbidden regions should proceed complementarily in terms of selected parameters for which the imaginary parts of precession angles are non - zero , i.e. for @xmath137 .
it is understood that this latter inequality proceeds in combination with @xmath138 .
such parameters belong to intervals like @xmath139 and @xmath140 which are complementary to ( 19 ) and ( 23 ) , respectively .
concrete plots displaying the parameter dependence of precession angles are presented in figs . 1 - 4 .
we have to anticipate that in all these cases the precession angles are positive whenever @xmath34 .
the precession angle is presented in fig . 1 for @xmath17 , @xmath34 ( solid curve ) and @xmath26 ( dashed curve ) .
the crossing points are located at @xmath141 , while the zeros are given by @xmath142 .
the admissible @xmath1-interval is given by @xmath143 , while the forbidden one proceeds via @xmath144 , in accord with the dotted curve in fig .
1 . the horizontal and vertical dotted lines serve to the discrimination of crossing points , and the same concerns figs . 2 - 4 . inserting @xmath145 instead of
@xmath17 leads to similar patterns , as shown in fig .
the crossing points and the zeros are now given by @xmath146 and @xmath147 , respectively .
the admissible interval concerns this time @xmath148 , which is complementary to the forbidden interval @xmath149 in which @xmath150 .
the @xmath0-dependence of the precession angle can be discussed in a rather similar manner .
inserting @xmath151 , leads to typical plots for @xmath34 ( solid curves ) and @xmath26 ( dashed curves ) , which exhibit the crossing points @xmath152 and the zeros @xmath153 , as shown in fig .
3 . however , this time the admissible interval is expressed by two disjoint semi - infinite intervals , i.e. by @xmath154 . the dotted curve concerns this time the forbidden interval @xmath155 .
the @xmath156- counterparts of these plots are displayed in fig .
4 . just note that in this latter case crossing points and zeros are given by @xmath157 and @xmath158 .
next let us rescale the squared wavenumber as @xmath159 which means that imaginary @xmath160realizations proceed via @xmath161 .
the @xmath23-dependence of @xmath162 can be easily displayed , as shown in fig .
5 for @xmath151 and @xmath17 .
one remarks that the imaginary parts of precession angles , such as indicated by dot - dashed ( @xmath34 ) and dotted ( @xmath26 ) curves , are zero , unless @xmath163 , where @xmath164 .
correspondingly , the real parts of present @xmath120-angles , which are displayed by solid ( @xmath34 ) and dashed ( @xmath26 ) curves , get characterized by a discontinuity point of second kind for which @xmath165 @xmath166 .
the imaginary and real parts obey the symmetry properties @xmath167 which is valid irrespective of @xmath23 and @xmath168 working for @xmath169 . a further discontinuity point of second kind is located at @xmath170 , such that @xmath171 , which is similar to the zeros displayed by @xmath172 in figs . 1 - 4 . a further interesting example concerns the @xmath0-dependence of real parts of precession angles for @xmath151 and @xmath173 , as shown by solid ( @xmath34 ) and dashed ( @xmath26 ) curves in fig . 6 .
dot - dashed ( @xmath34 ) and dotted ( @xmath26 ) curves , which are responsible for the imaginary parts , have again been inserted .
it is clear that now the rescaled wavenumber exhibits the complex form @xmath174 .
just remark the symmetry of present numerical realizations such as given by @xmath175 and @xmath176 vertical asymptotes reflecting the influence of crossing points @xmath177 discussed before by virtue of ( 23 ) can also be readily identified .
the present patterns are similar to the ones displayed in fig . 4 , but now one deals with additional imaginary contributions concerning @xmath178 as well as @xmath179 , this time for @xmath180 .
in this paper the parameter dependence of novel spin - precession effects proposed before@xcite has been discussed in some more detail . such effects concern equal coupling strength combinations of rashba and dresselhaus spin - orbit interactions characterized by the dimensionless coupling @xmath0 under the influence of in - plane magnetic fields of magnitude @xmath1 .
the zeeman interaction has been favoured to the detriment of the orbital effects of the magnetic field@xcite .
a related discrete parameter , say @xmath21 , has also accounted for .
the spin precession angle has been derived in terms of suitable @xmath35-expansions proceeding in terms of the convergence condition @xmath98 , which provides nontrivial manifestations .
this condition is fulfilled automatically when @xmath34 , but we have to resort to admissible regions in the parameter space such as given by ( 24 ) and ( 28 ) in so far as @xmath26 , i.e. to @xmath181 and @xmath182 .
to this aim parameter dependent crossing points between precession angles for @xmath34 and @xmath26 , namely @xmath183 @xmath184 and @xmath134 , have been established in an explicit manner .
in addition , the precession angles get characterized by nontrivial zeros like @xmath185 and @xmath186 , which are located specifically within admissible intervals .
it should be mentioned that the precession angles become imaginary within the forbidden intervals @xmath187 and @xmath188 , which are complementary to @xmath189 and @xmath190 , respectively .
meaningless vanishingly small strips starting upwards from @xmath191 and @xmath192 , respectively , have been reasonably ignored .
accordingly , the parameter dependence of the precession angles is characterized by the interplay between admissible and forbidden intervals , as illustrated by figs . 1 - 4 .
extrapolations of the wavenumber towards imaginary and complex values have also been done , as illustrated by figs . 5 and 6 .
such structures may be useful for further developments .
it is clear that starting from admissible intervals , the parameters can be tuned until the real part of the precession angle for @xmath26 becomes zero : @xmath193 this proceeds via @xmath194 @xmath187 or @xmath195 .
accordingly , the @xmath26-spin precession effects get ruled out .
this means that such effects are able to be switched off and on , which may serve to the description of further manipulations .
actual zeros of the precession angles can also be approached via @xmath196 and @xmath197 . in other words , one gets faced with nontrivial interplays between admissible and forbidden regions , which stand for the main results obtained in this paper .
so we are in a position to emphasize that such results are able to provide a deeper understanding of novel spin - precession effects characterizing quantum wire models . besides spin - filtering effects@xcite and transport properties@xcite ,
the incorporation of dynamic localization effects@xcite , as well as of time dependent magnetic fields@xcite , deserves further attention .
resorting to orthonormalized spinors like | in this paper one deals with the theoretical derivation of spin precession effects in quasi 1d quantum wire models .
such models get characterized by equal coupling strength superpositions of rashba and dresselhaus spin - orbit interactions of dimensionless magnitude @xmath0 under the influence of in - plane magnetic fields of magnitude @xmath1 . besides the wavenumber @xmath2 relying on the 1d electron , one accounts for the @xmath3 @xmath4 - factors in the front of the square root term of the energy .
electronic structure properties of quasi 1d semiconductor heterostructures like inas quantum wires can then be readily discussed .
indeed , resorting to the 2d rotation matrix provided by competing displacements working along the ox - axis opens the way to derive precession angles one looks for , as shown recently .
proceeding further , we have to resort reasonably to some extra conditions concerning the general selection of the @xmath2-wavenumber via @xmath5 , where @xmath6 stands for the nanometer length scale of the quantum wire
. we shall also account for rescaled wavenumbers , which opens the way to extrapolations towards imaginary and complex realizations .
the parameter dependence of the precession angles is characterized , in general , by interplays between admissible and forbidden regions , but large monotony intervals are also in order . |
nanoscale structures in the form of photonic and phononic crystals have recently been shown to feature significant radiation pressure interactions between localized optical cavity modes and internal nanomechanical resonances @xcite . alongside similar advances in the microwave domain @xcite , optomechanical crystals
have recently been used to laser cool a nanomechanical oscillator to its quantum ground state @xcite . the ability to measure and control the quantum state of such
an object ultimately hinges on the quantum efficiency of the optical transduction of motion .
here we demonstrate high - efficiency optical coupling between an optomechanical zipper cavity @xcite and a permanently mounted optical fiber through adiabatic mode conversion .
this optical coupling technique greatly improves the collection efficiency of light from these types of optomechanical cavities over existing methods , and brings the minimum total added measurement noise to within a factor of @xmath5 of the standard - quantum - limit of continuous position measurement . in a weak measurement of position through a parametrically
coupled optical cavity there are two intrinsic sources of measurement noise .
shot noise of the probe laser and excess quantum vacuum noise due to optical signal loss set the fundamental noise floor of the measurement .
when converted into units of mechanical quanta this imprecision noise @xmath6 decreases with increasing probe power .
however , higher laser probe power comes at the cost of radiation pressure backaction driving an additional occupation noise @xmath7 on top of the thermal mode occupation @xmath8 . for an optically resonant measurement of position , the noise terms in units of mechanical occupation quanta are @xmath9 where @xmath10 is the optomechanical interaction rate , @xmath11 and @xmath12 are respectively the optical and mechanical loss rates , @xmath13 is the extrinsic cavity loss rate , @xmath14 is the optical efficiency between the cavity and the detection channel , and @xmath15 accounts for excess technical noise and signal loss accumulated in experiment - specific optical components .
as intracavity photon number @xmath16 is varied , an optimal input power @xmath17 is reached where the imprecision noise and back - action noises are equal and the total added noise is minimized to @xmath18 in the ideal case this point , known as the standard - quantum - limit ( sql ) , adds @xmath19 quanta of noise to the measurement , equal to the zero - point fluctuations of the oscillator @xcite .
the sql can only be reached in the limit of noise - free , lossless detection ( @xmath20 ) and perfect waveguide loading ( @xmath21 ) .
thus , @xmath22 is a suitable figure of merit for the ultimate quantum efficiency of an optomechanical detector of position .
although experiments in both the optical and microwave domains have brought the imprecision noise level down to below @xmath23 quanta @xcite , and recent microwave experiments have acheived total added noise within a factor of @xmath24 of the sql @xcite , current state - of - the - art optical devices have been limited to @xmath25 times the sql @xcite .
such experiments are limited partly by technical noise ( e.g. added noise from amplifiers ) , but a substantial amount of imprecision is introduced by poor quantum efficiency of the optical readout . here
we identify another figure of merit to allow for cross - platform comparison of detection methods .
the quantum efficiency of a general measurement apparatus will be limited to @xmath26 , the collection efficiency of intracavity photons into the detection channel before further signal processing . to date
, most nanoscale optomechanical experiments utilize evanescent coupling between the optical cavity and an adiabatically tapered optical fiber @xcite .
while this method offers low parasitic losses , standing wave resonators such as the optomechanical cavities considered here radiate symmetrically into two oppositely propagating modes of the fiber , each at a rate @xmath13 .
thus the fraction of @xmath16 routed into the detection channel @xmath27 does not exceed @xmath19 even in the ideal case of negligible intrinsic cavity loss rate @xmath28 . to reach the overcoupled regime @xmath29 ,
a single - sided coupling scheme is necessary .
we have implemented a coupling scheme which routes light from a photonic crystal zipper cavity to a cleaved single - mode optical fiber tip with high efficiency in a fully single - sided interface .
the fiber is self - aligned in a si v - groove ( fig .
[ fig : fig_1 ] ) , secured with epoxy , and butt - coupled to a mode - matched si@xmath30n@xmath31 waveguide , which adiabatically widens @xcite to match the width of the zipper cavity nanobeam .
the waveguide then couples to the cavity through a truncated photonic crystal mirror .
the robust fiber alignment offers another key advantage over evanescent and grating - coupler techniques that call for nanometer - scale - sensitive positioning to achieve appreciable mode overlap .
nano - positioning is difficult and expensive to implement in cryogenic setups due to footprint and imaging requirements , whereas the coupler presented here can be installed directly into any system with a fiber port .
we now describe the optimization of the optical efficiency . to determine the optimal width of the si@xmath30n@xmath31 waveguide , we compute the guided transverse modes of both the waveguide and the optical fiber at the target wavelength of @xmath32 nm using a finite - element - method ( fem ) solver @xcite .
the coupling efficiency at the fiber - waveguide junction is calculated from the mode profiles using a mode overlap integral @xcite . for a @xmath33 nm thick si@xmath30n@xmath31 membrane , the optimal waveguide width is @xmath34 nm with a transmission efficiency of @xmath35 as depicted in fig .
[ fig : fig_1]c .
then @xmath36 increases gradually to the nanobeam width of @xmath37 nm . to obtain high transmission efficiency in this tapered waveguide section
, the rate of change of @xmath36 along the propagation direction @xmath38 must be small enough to satisfy the adiabatic condition @xcite @xmath39 at every point along the taper , where @xmath40 is the difference in effective index between the fundamental waveguide mode and any other guided or radiation mode .
the actual transmission efficiency is calculated using a finite - difference - time - domain ( fdtd ) simulation @xcite . for a @xmath41 m
long taper with a cubic shape between the junctions shown in fig .
[ fig : fig_1]c , d , we obtain a transmission efficiency of @xmath42 .
the tapered waveguide is supported by a @xmath43 nm wide tether placed near the fiber - waveguide junction ( fig .
[ fig : fig_1]c ) .
the scattering loss of the tether is computed using fdtd and the transmission efficiency is calculated to be @xmath44 . a second set of @xmath45 nm wide tethers ( fig . [ fig : fig_1]d ) is placed just before the cavity , in order to isolate the optomechanical crystal from the low - frequency vibrational modes of the tapered waveguide .
the waveguide is temporarily widened to @xmath46 m at this point , rendering the scattering loss due to the tethers negligible . finally , the uniform dielectric waveguide adiabatically transitions into a one - dimensional photonic crystal mirror by linearly increasing the radius of the holes while keeping the lattice constant fixed . an @xmath47 hole photonic crystal taper ( fig [ fig : fig_1]e ) is sufficient to acheive an efficiency of @xmath48 . the coupling rate to the cavity is controlled by varying the number of mirror periods after the taper . the photonic crystal cavity mode ( fig .
[ fig : fig_1]f ) is shared between the waveguide beam and a near - field , flexibly supported test beam , with the optomechanical coupling arising from the sensitivity of the resonance frequency to the beam separation . in this manner
the waveguide structure serves as an optical readout of the test beam motion with an optimal round - trip efficiency of @xmath49 . for the purposes of optomechanical transduction
, the relevant figure is the single - pass transmission efficiency between the cavity output and the fiber , which ideally is @xmath50 .
a broadband scan of the reflection from the fiber - coupled device is shown in fig . [
fig : fig_2]a . for probe wavelengths
detuned from cavity resonances , the photonic crystal depicted in fig .
[ fig : fig_1]d , e functions as a near - unity reflectivity mirror .
thus a low - finesse fabry - perot cavity is formed in the waveguide between the photonic crystal and the cleaved fiber facet ( with reflectivity @xmath51 ) of fig [ fig : fig_1]b . by fitting the visibility of the fringes to @xmath52 , the curve in fig .
[ fig : fig_2]a provides a convenient calibration of optical efficiency , which is plotted in fig .
[ fig : fig_2]b versus wavelength and reveals @xmath53 for resonant measurements of the primary mode at @xmath54 nm . the coupling depth of the mode is determined by fitting the resonance dip to a coupled - cavity model incorporating the photonic crystal and fabry - perot interference effects , yielding @xmath55 . to verify this model , we study a series of devices with varying numbers of mirror - type holes between the photonic crystal taper and cavity . in fig .
[ fig : fig_2]c , the total optical quality factor ( @xmath56 ) is plotted in green alongside intrinsic ( @xmath57 ) and extrinsic ( @xmath58 ) components as determined by the coupled cavity fit ( blue and red points respectively ) .
as the mirror holes are removed , @xmath58 decreases in good agreement with simulation ( dotted line ) , and the limiting component of @xmath56 transitions from @xmath57 to @xmath58 . by converting quality factor into loss rate and calculating the coupling depth of the device series , a clear trend from undercoupling to overcoupling is evident in fig .
[ fig : fig_2]d .
the device under test in this work features @xmath55 , bringing the fiber collection efficiency figure of merit to @xmath59 . for comparison , systems in which ground - state occupancy of a single mechanical mode has been achieved
have featured @xmath60 @xcite and @xmath61 @xcite .
to extract the noise power spectral density ( npsd ) of the test - beam displacement @xmath62 from the optomechanically phase - shifted cavity reflection , we place the device in an evacuated environment and connect the output fiber to the signal arm of a fiber - based balanced homodyne receiver shown schematically in fig [ fig : fig_3 ] .
we resonantly probe the optical resonance to avoid dynamic back - action on the test beam as well as to maximize the mechanical transduction sensitivity .
the probe laser frequency is positioned on resonance without frequency - locking to the cavity as drift is found to be negligibile over the time span of the measurement ( @xmath63 s ) .
the laser output is sent through a fiber polarization controller ( fpc ) and variable optical attenuator ( voa ) before being split by a variable beam splitter ( vbs1 ) into into the signal and local oscillator ( lo ) arms of the homodyne receiver .
the signal circulates through the device , while the lo propagates through a fiber stretcher ( fs ) before recombining with the signal at vbs2 , with the interference output detected on a balanced photodiode ( bpd ) pair .
the difference signal generated by the bpd is dediplexed at @xmath64 khz into a low - frequency lock signal ( fed back through a proportional - integral - derivative ( pid ) controller onto the fs to eliminate path - length fluctuations in the interferometer ) and a high - frequency photocurrent response .
the npsd of this high - frequency signal , @xmath65 , is measured on an electronic spectrum analyzer ( esa ) and plotted in fig .
[ fig : fig_4 ] for a probe power of @xmath66 nw . the optical efficiency of the setup to the device reflection signal is set by the transmission through the circulator ( @xmath67 ) and vbs2 ( @xmath68 ) as well as the bpd quantum efficiency ( @xmath69 ) , which combine to @xmath70 . the npsd shown in fig .
[ fig : fig_4 ] reveals a prominent resonance at @xmath71 mhz , which is identified using fem simulations as the fundamental in - plane mechanical mode of the test beam .
the optomechanical coupling rate of this mode , calibrated through the optical spring shift @xcite , is @xmath72 khz . a second prominent resonance exists at @xmath73 mhz which is identified as the fundamental in - plane mechanical mode of the waveguide with a lower optomechanical coupling ( @xmath74 khz ) owing to small overlap between the mechanical and optical modes .
overlays of the predicted single - sided transduction spectra , calculated using the measured parameters of the optomechanical cavity and fiber coupler , show good agreement with the measured signal for both the test - beam ( dashed green curve ) and the waveguide ( dashed red curve ) .
other peaks at @xmath5 mhz , @xmath75 mhz , and @xmath76 mhz do not correspond to any real mechanical motion of the beam but rather are due to nonlinear transduction of the mechanics .
thermal brownian motion of the beams , combined with the large optomechanical coupling , gives rise to a frequency shift of the optical mode which is a substantial fraction of the cavity optical linewidth , leading to harmonics in @xmath65 at multiples of the sum and difference frequencies of the two mechanical modes @xcite .
the remaining small features in @xmath65 are due to out - of - plane flexural modes of the structure , which are weakly transduced due to imperfect vertical symmetry .
for the remainder of the paper we focus our discussion on measurement of the mechanical mode with the strongest transduction , the fundamental in - plane mode of the test beam with mechanical frequency @xmath77 mhz .
the single - sided npsd of displacement for a harmonic mechanical mode at the mechanical resonance frequency is given by @xmath78 , where @xmath79 is the zero - point amplitude of the mechanical resonator with effective motional mass @xmath80 , and @xmath81 is the mode occupancy in units of phonon quanta . for the fundamental in - plane mode we numerically compute an effective mass of @xmath82 pg and a corresponding zero - point amplitude of @xmath83 fm .
when the mechanical mode is well - resolved in @xmath65 , that is , when the contributions of nearby mechanical modes are negligible and the resolution bandwidth of the esa is much less than @xmath12 , we can convert the spectrum into units of displacement for the measured mechanical mode by scaling @xmath84 to the computed value of @xmath85 .
we determine the imprecision in units of quanta for a measurement of this mechanical mode by referencing the measured background level to the height of the measured noise peak .
that is , the number of imprecision quanta is equal to @xmath86 divided by the signal - to - noise ratio ( snr ) of the resolved mechanical noise peak to the imprecision noise floor , so that an imprecision level of @xmath87 corresponds to the equivalent level of npsd that would be produced at the mechanical mode noise peak by a single quantum in the mechanical resonator @xcite . to determine the background noise floor we take the average of @xmath65 from @xmath88 mhz , indicated by relatively noise - free spectral region shaded gray in fig .
[ fig : fig_5]a . to directly measure the additional phonon occupation due to back - action
it is necessary to determine the total phonon occupancy , @xmath89 , by integrating over the full bandwidth of the mechanical mode .
however , such a study is outside the scope of the work presented here , as the large thermal occupation at room temperature ( @xmath90 ) dominates the signal over the comparatively small back - action , @xmath7 , produced by reasonable laser probe powers .
for now we restrict our attention to measuring the imprecision , and assume that the backaction is due only to the ideal radiation pressure term given in eqn .
[ eqn : noise ] .
this assumption is supported by measurements of the properties of known noise sources such as technical laser noise rad@xmath91 hz in the frequency range of interest .
consequently , for the probe powers used in this work , the excess back - action due to technical laser noise is negligible , and the phase noise contribution to the noise floor lies roughly @xmath92 db below the shot noise . ] , which is a common concern in optomechanical systems @xcite .
as there may exist additional , unknown sources of excess back - action in our measurement , we emphasize that the main result of this work is the collection efficiency of the coupling scheme , and that back - action and noise - driven occupation levels are used only to compare the ideal quantum limits of the device to the sql . by measuring the spectrum exemplified in fig [ fig : fig_4 ] for a range of probe powers
we observe the power dependence of @xmath6 , plotted in fig [ fig : fig_5]b with calculations of imprecision ( solid blue ) , ideal quantum back - action ( solid red ) , and total ( solid green ) noise quanta .
losses in the optical circuit set @xmath70 , which combined with @xmath93 and electronic noise brings the imprecision level of the measurement to @xmath3 times the ideal imprecision of a lossless device ( dashed blue curve ) .
the predicted total minimum added noise is thus @xmath94 phonons .
this minimum total added noise can be compared to similar measurements of mechanical position performed in the optical @xcite and microwave @xcite domains .
the coupling scheme here compares favorably to the minimum total added noise of @xmath95 demonstrated with whispering - gallery - mode resonators @xcite , @xmath96 demonstrated with membranes in fabry - perot cavities @xcite , and @xmath97 demonstrated with microwave resonators @xcite .
alternatively , one can compare to state - of - the - art ground state cooling experiments @xcite where the occupation of the mechanical resonator is brought near or below a single phonon and the quantum limits of the mechanical measurement become crucial .
in such experiments the back - action is used to passively cool the mechanical resonator , and thus the relevant noise term is simply @xmath6 . in the red - detuned ,
sideband resolved regime relevant for ground - state cooling , the imprecision can not be made arbitrarily small by increasing the power , but rather has the asymptotic form @xmath98 in the limit of large intracavity photon number .
our coupling scheme allows us to achieve @xmath99 , which lies well below the imprecision level previously demonstrated using similar nanoscale optomechanical cavties ( @xmath100 ) @xcite , and is comparable to the imprecision acheived in whispering - gallery - mode resonators ( @xmath101 ) @xcite and microwave resonators ( @xmath102 ) @xcite .
it is worth noting that improvements to @xmath15 of the optical circuit such as higher efficiency photodiodes can lower @xmath22 to below @xmath103 quantum .
in addition to improving the collection efficiency towards the theoretical maximum of @xmath104 , further work will focus on measuring the back - action exerted on the resonator by placing the device in suitable cryogenic conditions where the thermal occupation is strongly suppressed .
another key application which benefits greatly from the improved collection efficiency demonstrated here is feedback damping of the mechanical motion @xcite , which is fundamentally limited by the imprecision noise and could enable ground state cooling of sideband unresolved mechanical systems such as the zipper cavity .
this work was supported by the darpa / mto orchid and meso programs , the institute for quantum information and matter , an nsf physics frontiers center with support of the gordon and betty moore foundation , and by the afosr qumpass muri .
we gratefully acknowledge critical support and infrastructure provided for this work by the kavli nanoscience institute at caltech . | a significant challenge in the development of chip - scale cavity - optomechanical devices as testbeds for quantum experiments and classical metrology lies in the coupling of light from nanoscale optical mode volumes to conventional optical components such as lenses and fibers . in this work
we demonstrate a high - efficiency , single - sided fiber - optic coupling platform for optomechanical cavities . by utilizing an adiabatic waveguide taper to transform a single optical mode between a photonic crystal zipper cavity and a permanently mounted fiber ,
we achieve a collection efficiency for intracavity photons of @xmath0 at the cavity resonance wavelength of @xmath1 nm .
an optical balanced homodyne measurement of the displacement fluctuations of the fundamental in - plane mechanical resonance at @xmath2 mhz reveals that the imprecision noise floor lies a factor of @xmath3 above the standard quantum limit ( sql ) for continuous position measurement , with a predicted total added noise of @xmath4 phonons at the optimal probe power .
the combination of extremely low measurement noise and robust fiber alignment presents significant progress towards single - phonon sensitivity for these sorts of integrated micro - optomechanical cavities .
10 url # 1#1urlprefix m. eichenfield , j. chan , r. m. camacho , k. j. vahala , and o. painter , `` optomechanical crystals , '' nature * 462 * , 7882 ( 2009 ) .
j. teufel , t. donner , d. li , j. harlow , m. allman , k. cicak , a. siroi , j. whittaker , k. lehnert , and r. simmonds , `` sideband cooling of micromechanical motion to the quantum ground state , '' nature * 475 * , 359363 ( 2011 ) .
j. chan , t. mayer - alegre , a. h. safavi - naeini , j. t. hill , a. krause , s. grblacher , m. aspelmeyer , and o. painter , `` laser cooling of a nanomechanical oscillator into its quantum ground state , '' nature * 478 * , 8992 ( 2011 ) .
m. eichenfield , r. camacho , j. chan , k. j. vahala , and o. j. painter , `` a picogram- and nanometre - scale photonic - crystal optomechanical cavity , '' nature * 459 * , 550555 ( 2009 ) .
a. a. clerk , m. h. devoret , s. m. girvin , f. marquardt , and r. j. schoelkopf , `` introduction to quantum noise , measurement and amplification , '' review of modern physics * 82 * , 11551208 ( 2010 ) .
j. teufel , t. donner , m. castellanos - beltran , j. harlow , and k. lehnert , `` nanomechanical motion measured with an imprecision below that at the standard quantum limit , '' nature nanotechnology * 4 * , 820823 ( 2009 ) .
g. anetsberger , e. gavartin , o. arcizet , q. unterreithmeier , e. weig , m. gorodetsky , j. kotthaus , and t. kippenberg , `` measuring nanomechanical motion with an imprecision below the standard quantum limit , '' physical review a * 82 * , 061804 ( 2010 ) .
r. rivire , s. delglise , s. weis , e. gavartin , o. arcizet , a. schliesser , and t. kippenberg , `` optomechanical sideband cooling of a micromechanical oscillator close to the quantum ground state , '' physical review a * 83 * , 063835 ( 2011 ) . c. michael , m. borselli , t. johnson , c. chrystal , and o. painter , `` an optical fiber - taper probe for wafer - scale microphotonic device characterization , '' optics express * 15 * , 47454752 ( 2007 ) .
s. spillane , t. kippenberg , o. painter , and k. vahala , `` ideality in a fiber - taper - coupled microresonator system for application to cavity quantum electrodynamics , '' physical review letters * 91 * , 043902 ( 2003 ) .
o. mitomi , k. kasaya , and h. miyazawa , `` design of a single - mode tapered waveguide for low - loss chip - to - fiber coupling , '' ieee journal of quantum electronics * 30 * , 17871793 ( 1994 ) .
v. r. almeida , r. r. panepucci , and m. lipson , `` nanotaper for compact mode conversion , '' optics letters * 28 * , 13021304 ( 2003 ) .
s. j. mcnab , n. moll , and y. a. vlasov , `` ultra - low loss photonic integrated circuit with membrane - type photonic crystal waveguids , '' optics express * 11 * , 29272939 ( 2003 ) .
l. chen , c. r. doerr , y .- k .
chen , and t .- y .
liow , `` low - loss and broadband cantilever couplers between standard cleaved fibers and high - index - contrast si3n4 or si waveguides , '' ieee photonics technology letters * 22 * , 17441746 ( 2010 ) .
comsol multiphysics http://www.comsol.com/ .
a. w. snyder and j. d. love , _ optical waveguide theory _
( springer , 1983 ) .
lumerical solutions inc .
http://www.lumerical.com/tcad-products/fdtd/ .
q. lin , j. rosenberg , x. jiang , k. vahala , and o. painter , `` mechanical oscillation and cooling actuated by the optical gradient force , '' physical review letters * 103 * , 10360 ( 2009 ) .
p. rabl , c. genes , k. hammerer , and m. aspelmeyer , `` phase - noise induced limitations on cooling and coherent evolution in optomechanical systems , '' physical review a * 80 * , 063819 ( 2009 ) . a. h. safavi - naeini , j. chan , j. t. hill , s. grblacher , h. miao , y. chen , m. aspelmeyer , and o. painter , `` laser noise in cavity - optomechanical cooling and thermometry , '' arxiv:1210.2671v1 .
a. m. jayich , j. c. sankey , k. bjorke , d. lee , c. yang , m. underwood , l. childress , a. petrenko , s. m. girvin , and j. g. e. harris , `` cryogenic optomechanics with a si3n4 membrane and classical laser noise , '' new journal of physics * 14 * , 115018 ( 2012 ) .
t. p. purdy , r. w. peterson , and c. a. regal , `` observation of radiation pressure shot noise , '' arxiv:1209.6334v1 .
c. genes , d. vitali , p. tombesi , s. gigan , and m. aspelmeyer , `` ground - state cooling of a micromechanical oscillator : comparing cold damping and cavity - assisted cooling schemes , '' physical review a * 77 * , 033804 ( 2008 ) .
k. d. and d. bouwmeester , `` sub - kelvin optical cooling of a micromechanical resonator , '' nature * 444 * , 7578 ( 2006 ) . |
vbfnlo @xcite is a flexible monte carlo ( mc ) program for vector boson fusion ( vbf ) , double and triple vector boson ( plus jet ) production processes at nlo qcd accuracy .
the electroweak corrections to higgs boson production via vbf have been included .
in addition , the simulation of @xmath16-even and @xmath16-odd higgs boson production in gluon fusion , associated with two additional jets , is implemented at the ( one - loop ) lo qcd level .
vbfnlo can be run in the mssm , and anomalous couplings of the higgs boson and gauge bosons have been implemented for certain processes .
additionally , two higgsless extra dimension models are included the warped higgsless scenario and a three - site higgsless model for selected processes .
further information , and the latest version of the code , can be found on the vbfnlo webpage http://www-itp.particle.uni-karlsruhe.de/vbfnlo/ a complete process list is given in appendix a.
the latest version of vbfnlo has several new processes implemented at nlo qcd . same sign @xmath5 pair production with two jets via
vector boson fusion @xcite has been included in vbfnlo 2.6.0 .
this process is potentially sensitive to new physics signals and , as it gives rise to same sign dilepton final states , it is also a background to new physics scenarios . to distinguish potential signatures of physics beyond the standard model from the effect of higher order corrections , precise theoretical predictions are essential .
although the qcd corrections to the integrated cross sections were found to be relatively small , their effect on several distributions is appreciable . when nlo qcd corrections are taken into account , the residual scale uncertainties are at the 2.5% level .
the new process ids are given in table [ tab : vvjj ] . | vbfnlo is a flexible parton level monte carlo program for the simulation of vector boson fusion ( vbf ) , double and triple vector boson ( plus jet ) production in hadronic collisions at next - to - leading order ( nlo ) in the strong coupling constant , as well as higgs boson plus two jet production via gluon fusion at the one - loop level .
this note briefly describes the main additional features and processes that have been added in the new release vbfnlo version 2.6.0 . at nlo qcd diboson production ( @xmath0 , @xmath1 , @xmath2 , @xmath3 and @xmath4 ) , same - sign @xmath5 pair production via vector boson fusion and the process
@xmath6 have been implemented ( for which one - loop tensor integrals up to six - point functions are included ) .
in addition , gluon induced diboson production can be studied separately at the leading order ( one - loop ) level .
the diboson processes @xmath7 , @xmath1 and @xmath0 can be run with anomalous gauge boson couplings , and anomalous couplings between a higgs and a pair of gauge bosons is included in @xmath7 , @xmath2 , @xmath3 and @xmath4 diboson production .
the code has also been extended to include anomalous gauge boson couplings for single vector boson production via vbf , and a spin-2 model has been implemented for diboson pair production via vector boson fusion .
ka - tp-28 - 2012 , man / hep/2012/09 , mz - th/12 - 27 , + cern - ph - th/2012 - 194 , ippp/12/50 , dcpt/12/100 , + ftuv-12 - 0709 , lpn12 - 078 , ttk-12 - 32 * release note vbfnlo-2.6.0 * k. arnold@xmath8 , j. bellm@xmath8 , g. bozzi@xmath9 , f. campanario@xmath8 , c. englert@xmath10 , b. feigl@xmath8 , j. frank@xmath8 , t. figy@xmath11 , b. jger@xmath12 , m. kerner@xmath8 , m. kubocz@xmath13 , c. oleari@xmath14 , s. palmer@xmath8 , m. rauch@xmath8 , h. rzehak@xmath15 , f. schissler@xmath8 , o. schlimpert@xmath8 , m. spannowsky@xmath10 , d. zeppenfeld@xmath8 @xmath8 institut fr theoretische physik , universitt karlsruhe , karlsruhe institute of technology , 76128 karlsruhe , germany + @xmath9 dipartimento di fisica , universita di milano and infn , 20133 milano , italy + @xmath10 institute for particle physics phenomenology , university of durham , durham , dh1 3le , united kingdom + @xmath11 school of physics and astronomy , the university of manchester , manchester , m13 9pl , united kingdom + @xmath12 institut fr physik ( thep ) , johannes - gutenberg - universitt , 55099 mainz , germany + @xmath13 institut fr theoretische teilchenphysik und kosmologie , rwth aachen university , 52056 aachen , germany + @xmath14 dipartimento di fisica , universit di milano - bicocca and infn , sezione di milano - bicocca , 20126 milano , italy + @xmath15 cern , ch-1211 geneva 23 , switzerland ( on leave from : physikalisches institut albert - ludwigs - universitt freiburg , hermann - herder - str . 3 , 79104 freiburg i m breisgau , germany ) + |
in multiple input and multiple output ( mimo ) systems , channel adaptive techniques ( e.g. , water - filling , interference alignment , beamforming , etc . ) can enhance the spectral efficiency or the capacity of the system .
however , these channel adaptive techniques require accurate channel conditions , often referred to channel state information ( csi ) . oftentimes , in a frequency - division duplex ( fdd ) setting , csi is estimated at the receiver and conveyed to the transmitter via a feedback channel . in recent years
, csi feedback problems have been intensively studied , due to its potential benefits to the mimo systems @xcite .
it is significant to explore how to reduce the feedback load , due to the uplink feedback channel limitation . in @xcite ,
four feedback rate reduction approaches were reviewed , where the lossy compression using the properties of the fading process was considered best .
when the wireless channel experiences temporal - correlated fading , modeled as a finite - state markov chain , the amount of csi feedback bits can be reduced by ignoring the states occurring with small probabilities @xcite .
the feedback rate in frequency - selective fading channels was studied in @xcite , by exploiting the frequency correlation . in summary , all the above works mainly focus on feedback rate compression considering either temporal correlation or spectral correlation .
however , doubly selective fading channels are more frequently encountered in wireless communications as the desired data rate and mobility grow simultaneously . to the best knowledge of the authors , the scheme of making full use of the two - dimensional correlations
is not yet well studied . using both of the orthogonal dimensional correlations in a cooperated way
, the feedback overhead can be further reduced in the doubly selective fading channels .
thus , in this paper , we derive the minimal feedback rate using both the temporal and spectral correlations .
the main contributions of the present paper can be briefly summarized as:1 ) we discuss the minimal feedback rate without differential feedback .
2 ) we propose a differential feedback scheme by exploiting the temporal and spectral correlations , and 3 ) we derive the minimal differential feedback rate expression over mimo doubly selective fading channel . the rest of the paper is organized as follows : in section ii , we describe the differential feedback model as well as the statistics of the doubly selective fading channel . in section iii ,
we propose a differential feedback scheme by exploiting the two - dimensional correlations and derive the minimal feedback rate . in section
iv , we provide some simulation results showing the performance of the proposed scheme .
in this paper , we assume that the down - link channel is a mobile wireless channel which is always correlated in time and frequency domains , while the up - link channel is a limited feedback channel . since the channel corresponding to each antenna is independent and with the same statistics , we can describe the separation property of the channel frequency response @xmath0 at time @xmath1 for an arbitrary transmit and receive antenna pair @xcite @xmath2 where @xmath3 denotes expectation function , the superscript @xmath4 denotes complex conjugate .
@xmath5 is the power of the channel frequency response . @xmath6 and @xmath7 denotes the temporal and spectral correlation functions , respectively . assuming that the channel frequency response stays constant within the symbol period @xmath8 and the subchannel spacing @xmath9 , the correlation function for different periods and subchannels is written as @xmath10 = \sigma_h^2 { r_t}\left [ \delta m \right]{r_f}\left[\delta
n \right],\ ] ] where @xmath11 = { \kern 1pt } { r_t}\left ( { \delta m{t_s } } \right)$ ] and @xmath12{\kern 1pt } { \kern 1pt } = r\left ( { \delta n{f_s } } \right)$ ] .
furthermore , if we just consider the time domain , the correlated channel can be modeled as a time - domain first - order autoregressive process ( ar1)@xcite@xmath13 where @xmath14 denotes the channel coefficient of the @xmath15th symbol interval and the @xmath16th subchannel , @xmath17 is a complex white noise variable , which is independent of @xmath18 , with variance @xmath19 .
the parameter @xmath20 is the time autocorrelation coefficient , which is given by the zero - order bessel function of first kind @xmath21 = { j_0}\left ( { 2\pi { f_d}t_s } \right)$ ] , where @xmath22 is the doppler frequency @xcite .
similarly , if we just consider the frequency domain , the correlated channel can also be represented as a frequency - domain ar1 @xcite @xmath23 where @xmath24 is a complex white noise variable , which is independent of @xmath25 , with variance @xmath19 .
the parameter @xmath26 determines the correlation between the subchannels , which is given by @xmath27=\frac{1}{\sqrt{1+(2\pi f_s \delta ) ^2}}$ ] , where @xmath28 is the root mean square delay spread @xcite . the system model with differential feedback
is illustrated in fig .
[ fig : sm ] . by using differential feedback scheme
, the receiver just feeds back the differential csi .
we suppose that there are @xmath29 and @xmath30 antennas at the transmitter and receiver , respectively . the received signal vector at the @xmath15th symbol interval and the @xmath16th subchannel is given by @xmath31 in the above expression , @xmath32 denotes the @xmath33 received vector at the @xmath15th symbol interval and the @xmath16th subchannel .
@xmath34 , a @xmath35 channel fading matrix , is the frequency response of the channel .
the entries are assumed independent and identically distributed ( i.i.d . ) , obeying a complex gaussian distribution with zero - mean and variance @xmath19 .
different antennas have the same characteristic in temporal and spectral correlations , @xmath20 and @xmath26 , respectively . besides , there is no spatial correlation between different antennas .
@xmath36 denotes the @xmath37 transmitter signal vector and is assumed to have unit variance .
@xmath38 is a @xmath39 additive white gaussian noise ( awgn ) vector with zero - mean and variance @xmath40 .
both @xmath36 and @xmath38 are independent for different @xmath15 s and @xmath16 s . through csi quantization , the feedback channel output is written as @xcite @xmath41 where @xmath42 denotes the channel quantization matrix , and @xmath43 is the independent additive quantization distortion matrix whose entries are zero - mean and with variance @xmath44 , where @xmath45 represents the channel quantization distortion constraint .
the differential feedback is under consideration as shown in fig .
[ fig : sm ] .
we can use the previous csi to forecast the present csi @xmath46 at the transmitter @xmath47 where @xmath48 and @xmath49 are the coefficients of the channel predictor which will be calculated by using the minimum mean square error ( mmse ) principle in the next section .
meanwhile , the receiver calculates the differential csi , given the previous ones .
the differential csi can be formulated as @xmath50 where @xmath51 represents the differential csi which obviously is the prediction error , and @xmath52 is the differential function .
then through limited feedback channel , @xmath53 should be quantized and fed back .
finally , the csi reconstructed by combining the differential one and the channel prediction is utilized by the channel adaptive techniques . in this paper , we adopt the water - filling precoder , however , the analysis and conclusions given in this paper are also valid for other adaptive techniques .
the channel quantization matrix is decomposed as @xmath54 using singular value decomposition ( svd ) at the transmitter . @xmath55 and @xmath56 are unitary matrixes , and @xmath57 is a non - negative diagonal matrix composed of eigenvalues of @xmath58 . with the water - filling precoder , the closed - loop capacity can be obtained as @xcite @xmath59,\ ] ] where @xmath60 , @xmath61 , and @xmath62 $ ] , where @xmath63 represents the amplitude of signal symbol , and @xmath64 denotes a diagonal matrix determined by water - filling @xcite @xmath65 where @xmath66 are the entries of @xmath67 , @xmath68 denotes a cut - off value chosen to meet the power constraint .
it is obvious from ( [ eq : sm_erg_capacity ] ) that the closed - loop ergodic capacity is determined by @xmath46 and @xmath69 , and the loss of capacity is mainly caused by the quantization error .
therefore , given the limited feedback channel , the capacity can be enhanced by exploiting the channel correlations to reduce the quantization error .
in this section , exploiting the temporal and spectral correlations , we study the minimal feedback rate that denotes the minimal feedback bits required per block to preserve the given channel quantization distortion .
we first describe the feedback rate using normal quantization . without differential feedback scheme ,
the receiver feeds back @xmath70 to the transmitter .
the information entropy of a gaussian variable @xmath71 with variance @xmath72 is represented as @xcite @xmath73 thus , the feedback load has positive relation with @xmath19 .
furthermore , taking quantization of the channel matrix into consideration , the feedback rate is determined by the rate distortion theory of continuous - amplitude sources @xcite @xmath74 \le d } \right\},\ ] ] where @xmath75 denotes infimum function , @xmath76 denotes the mutual information between @xmath77 and @xmath78 , and @xmath79 denotes the channel quantization distortion which is constrained by @xmath45 . since the entries of @xmath80 and @xmath81 are i.i.d .
complex gaussian variables , the feedback rate can be written as @xmath82 \le d } \right\},\ ] ] where @xmath83 is the one - dimensional average channel quantization distortion . @xmath84 and @xmath85 represent the entries of @xmath86 , @xmath87 , respectively .
also , from ( [ eq : sm_quan_channel ] ) the one - dimensional channel quantization is written as @xmath88 the mutual information can be written as @xmath89 combining ( [ eq : mfr_1d_quan ] ) , ( [ eq : mfr_mutl_info ] ) can be rewritten as @xmath90 substituting ( [ eq : mfr_info_entrpy ] ) and ( [ eq : mfr_mutl_info1 ] ) into ( [ eq : mfr_1d_minfb ] ) , we obtain @xmath91 from ( [ eq : mfr_no_finalrate ] ) , the feedback rate required for the non - differential feedback is very large . nevertheless , by employing the temporal and spectral correlations , we can use the differential feedback scheme to reduce the feedback bits significantly .
the transmitter can predict the present csi @xmath92 depending on the previous ones in time domain @xmath93 and frequency domain @xmath93 . then
, the receiver quantizes @xmath51 , or equivalently , the error of the channel prediction , and feeds back to the transmitter .
finally , the transmitter reconstructs the csi by both the channel prediction and the differential csi .
it is obvious that the more accurate the channel is predicted , the less bits is fed back from the receiver . as @xmath94 and @xmath34 are correlated , an mmse channel predictor can be constructed as ( [ eq : sm_channel_esti ] ) , where the coefficients @xmath48 and @xmath49 are selected to minimize @xmath95 the mse represents the statistical difference between the predicted value and the true one .
we can obtain the minimized quantization bits by minimizing the mse .
we can rewrite @xmath86 as @xmath96 where @xmath97 is the differential feedback load to minimize . by the orthogonality principle @xcite ,
@xmath98 are determined by @xmath99 = 0\\ \mathbb{e}\left [ { \left ( { { { \bf{h}}_{m , n } } - { a_1}{{\bf{h}}_{m - 1,n } } - { a_2}{{\bf{h}}_{m , n - 1 } } } \right){{\bf{h}}_{m , n - 1 } } } \right ] = 0 \end{array } \right .. \ ] ] since the entries of @xmath70,@xmath100,@xmath101 are i.i.d .
complex gaussian variables , the orthogonality principle can be rewritten as @xmath102 = 0\\ \mathbb{e}\left [ { \left ( { { h_{m , n } } - { a_1}{h_{m - 1,n } } - { a_2}{h_{m , n - 1 } } } \right){h_{m , n - 1 } } } \right ] = 0 \end{array } \right .. \ ] ] moreover , the one - dimensional frequency response of the channel can be represented as @xmath103 where @xmath84 , @xmath104 , @xmath105 , @xmath106 and @xmath107 represent the corresponding entries .
direct calculation shows that ( [ eq : mfr_orth_principle_single ] ) is equivalent to @xmath108 - { a_1}{r_h}\left [ { 0,0 } \right ] - { a_2}{r_h}\left [ { 1,1 } \right ] = 0\\ { r_h}\left [ { 0,1 } \right ] - { a_1}{r_h}\left [ { 1,1 } \right ] - { a_2}{r_h}\left [ { 0,0 } \right ] = 0 \end{array } \right .. \ ] ] with the separation property of the correlations of the channel frequency response ( [ eq : sm_discrete_correlation_function ] ) , and combining @xmath109={r_f}\left [ 0 \right]=1 $ ] and @xmath110=\alpha_t$ ] , @xmath111=\alpha_f$ ] , ( [ eq : mfr_orth_prin_direct1 ] ) can be simplified by @xmath112 from ( [ eq : mfr_orth_prin_substitue ] ) , @xmath113 are given by @xmath114 combing ( [ eq : mfr_a1_a2 ] ) and ( [ eq : mfr_1d_hmn ] ) , the one - dimensional mse of the channel estimator is @xmath115 finally , the channel estimator @xmath116 is given by @xmath117 and combining ( [ eq : mfr_hmn_rewrite ] ) and ( [ eq : mfr_chan_est_final ] ) , @xmath86 is given by @xmath118 then , through the feedback channel , the error of the channel predictor @xmath119 can be fed back from the transmitter to the receiver .
similarly , from ( [ eq : mfr_info_entrpy ] ) , the feedback load is positive related with @xmath120 . because @xmath121 , @xmath122 , the feedback load can be much smaller than @xmath19 , the non - differential one , especially when the channel is highly correlated .
for example , given @xmath123 , @xmath124 , then @xmath125 . from ( [ eq : mfr_hmn_fianl_expression ] ) ,
taking quantization impact into consideration , the minimal differential feedback rate over doubly selective fading channels can be calculated by the rate distortion theory of continuous - amplitude sources in a similar way . @xmath126 where the channel predictor coefficients @xmath98 are determined by @xmath127 and @xmath128 .
the average power of @xmath129 is @xmath130 .
the detailed derivation is given in appendix a. the above expression gives the minimal differential feedback rate simultaneously utilizing the temporal and spectral correlations . from ( [ eq : mfr_minfb_tf ] ) , the minimal differential feedback rate is a function of @xmath131 and the channel quantization distortion @xmath132 , and much smaller than that of the non - differential one ( [ eq : mfr_no_finalrate ] ) .
in this section , we first provide the relationship between the mse of the predictor and the two - dimensional correlations in fig .
[ fig : mse ] . the minimal differential feedback rate over mimo doubly selective fading channels
is given in fig .
[ fig : ft_mini_fb ] .
then , a longitudinal section of fig . [ fig : ft_mini_fb ] is presented , where we assume the temporal correlation and spectral correlation is equal .
finally , we verify our theoretical results by a practical differential feedback system with water - filling precoder and lloyd s quantization algorithm @xcite . for simplicity and without loss of generality
, we consider @xmath133 , and @xmath134 . fig .
[ fig : mse ] presents the mse between the predicted value and the true value . as the temporal or spectral correlation increases ,
the mse decreases .
furthermore , when either @xmath20 or @xmath26 comes to one , the mse tends to zero . and
@xmath135.,width=316 ] fig .
[ fig : ft_mini_fb ] plots the relationship between the minimal differential feedback rate and the two - dimensional correlations with the channel quantization distortion @xmath135 .
it is very similar to the mse shown in fig .
[ fig : mse ] , because it presents the minimal bits required to quantize the differential csi . and @xmath135.,width=316 ] additionally ,
because @xmath20 and @xmath26 could be any value , we provide one of the longitudinal section of fig .
[ fig : ft_mini_fb ] where the temporal correlation is equal to the spectral correlation in fig .
[ fig : sect_mini_fb ] . for comparison ,
the differential feedback compression only using one - dimensional correlation and the non - differential feedback scheme are also included in fig . [ fig : sect_mini_fb ] .
it is observed from fig .
[ fig : sect_mini_fb ] that the scheme using both temporal and spectral correlations is always better than the scheme using only one - dimensional correlation . as the correlations increase ,
the two - dimensional differential feedback compression exhibits a significant improvement compared to one - dimensional one .
this performance advantage even reaches up to @xmath136 with @xmath137 . and
@xmath135.,width=316 ] in this subsection , we consider the temporal correlation @xmath138 , with carrier frequency @xmath139 ghz , the normalized doppler shift @xmath140 hz and spectral correlation @xmath141 , with @xmath142 , which is a reasonable assumption @xcite .
we design a differential feedback system using lloyd s quantization algorithm to verify our theoretical results @xcite .
we use @xmath143 as a differential function , where @xmath144 , @xmath145 in the two - dimensional differential feedback compression and @xmath146 , @xmath147 in the one - dimensional one .
the feedback steps can be summarized as follows .
firstly , based on lloyd s quantization algorithm , the channel codebook can be generated according to the statistics of the corresponding differential feedback load at both transmitter and receiver .
secondly , the receiver calculates the current differential csi @xmath53 .
thirdly , the differential csi is quantized to the optimal coodbook value @xmath148 according to the euclidean distance .
finally , the transmitter reconstructs the channel quantization matrix by @xmath149 . in fig .
[ fig : cap_t_f ] , we give the simulation results of the ergodic capacity employing lloyd s algorithm .
the theoretical capacity results are also provided in fig .
[ fig : cap_t_f ] .
we can see from fig.[fig : cap_t_f ] that the performance of the two - dimensional one are always better than the one - dimensional one , which verifies our theoretical analysis . and
snr @xmath150db.,width=316 ] as shown in fig.[fig : cap_t_f ] , with the increase of feedback rate @xmath151 , the ergodic capacities increase rapidly when @xmath151 is small , and then slow down in the large @xmath151 region , because when @xmath151 is large enough , the quantization errors tend to zero . also , the capacities of lloyd s quantization are lower than the theoretical ones .
the reasons are as follows .
the lloyd s algorithm is optimal only in the sense of minimizing a variable s quantization error , but not in data sequence compression while the channel coefficient @xmath152 is correlated in both temporal and spectral domain .
however , the imperfection reduces as @xmath151 increases , because the quantization errors of both lloyd s algorithm and theoretical results tend to zero with sufficient feedback bits @xmath151 .
in this paper , we have designed a differential feedback scheme making full use of both the temporal and spectral correlation and compared the performance with the scheme without differential feedback .
we have derived the minimal differential feedback rate for our proposed scheme .
the feedback rate to preserve the given channel quantization distortion is significantly small compared to non - differential one , as the channel is highly correlated in both temporal and spectral domain .
finally , we provide simulations to verify our analysis .
the minimal differential feedback rate over mimo doubly selective fading channel can also be derived by the rate distortion theory . given @xmath153 and @xmath154 at the transmitter , the differential feedback rate can be represented as @xmath155\hspace{-1.2 mm } \le\hspace{-1.2 mm } d } \hspace{-0.6mm}\right\}\hspace{-0.8mm}.\ ] ] since the entries are i.i.d .
complex gaussian variables , ( [ eq : apdixb_r_inf ] ) can be written as @xmath156\hspace{-1.2 mm } \le\hspace{-1.2 mm } d } \hspace{-0.6mm}\right\}\hspace{-0.8mm}.\ ] ] the one - dimensional channel quantization equality can be written as @xmath157 similarly , ( [ eq : mfr_hmn_fianl_expression ] ) yields @xmath158 where @xmath159 , @xmath160 .
the conditional mutual information @xmath161 can be written as @xmath162 first , we calculate @xmath163 . substituting ( [ eq : apdixb_1d_quan ] ) into ( [ eq : apdixb_1d_hmn ] )
, it yields that @xmath164 substituting ( [ eq : apdixb_1d_hmnquan ] ) into ( [ eq : apdixb_1d_i ] ) , we obtain @xmath165 considering inequality @xmath166 ( [ eq : apdixb_1d_i_cal1 ] ) can be written as @xmath167 since @xmath168 , @xmath169 and @xmath129 are complex gaussian variables , and the information entropy of a gaussian variables with variance @xmath72 is @xmath170 , we calculate the variance of @xmath171 @xmath172 now we give the derivation of the correlation function of two noise terms @xmath173 . from ( [ eq : apdixb_1d_quan ] )
, the quantization error can be decomposed into two parts @xmath174 where @xmath175 @xmath176 is a gaussian variable with zero - mean and variance @xmath177 , independent with @xmath178 .
y. zhang , r. yu , m. nekovee , y. liu , s. xie , and s. gjessing , `` cognitive machine - to - machine communications : visions and potentials for the smart grid '' , _ ieee network magazine _ ,
vol.26 , no.3 , pp.613 , may / jun . 2012 .
w. h. chin and c. yuen , design of differential quantization for low bitrate channel state information feedback in mimo - ofdm systems , " in _ ieee vtc - spring 2008_. y. li , l. j. cimini , jr . , and n. r. sollenberger , `` robust channel estimation for ofdm systems with rapid dispersive fading channels , '' _ ieee trans .
46 , no . 7 , pp
. 902915 , jul .
1998 .
z. shi , s. hong , j. chen , k. chen and y. sun , `` particle filter based synchronization of chaotic colpitts circuits combating awgn channel distortion '' , _ circuits , systems and signal processing ( springer ) _ ,
vol.27(6 ) : 833845 , dec . 2008 .
l. zhang , l. song , m. ma , and b. jiao , `` on the minimum differential feedback rate for time - correlated mimo rayleigh block - fading channels , '' _ ieee trans .
2 , pp . 411420 , feb . 2012 . | channel state information ( csi ) provided by limited feedback channel can be utilized to increase the system throughput . however , in multiple input multiple output ( mimo ) systems , the signaling overhead realizing this csi feedback can be quite large , while the capacity of the uplink feedback channel is typically limited .
hence , it is crucial to reduce the amount of feedback bits .
prior work on limited feedback compression commonly adopted the block fading channel model where only temporal or spectral correlation in wireless channel is considered . in this paper , we propose a differential feedback scheme with full use of the temporal and spectral correlations to reduce the feedback load .
then , the minimal differential feedback rate over mimo time - frequency ( or doubly ) selective fading channel is investigated .
finally , the analysis is verified by simulation results .
differential feedback , correlation , mimo |
there are two types of quantum stochastic differential equations , namely , linear and nonlinear .
linear quantum systems can be described by linear quantum stochastic differential equations and arise mostly in the area of quantum optics ; e.g. , see @xcite , @xcite and @xcite . an important class of linear quantum stochastic models describe the heisenberg evolution of the annihilation and creation operators of several independent open quantum harmonic oscillators that are coupled to external coherent bosonic fields , such as coherent laser beams ; e.g. , see @xcite , @xcite and @xcite . a special class of these linear quantum systems is driven by quantum wiener processes as in @xcite where physical realizability conditions are developed to determine when the linear quantum system under consideration can be regarded as a representation of a linear quantum harmonic oscillator .
this imposes some restrictions on the matrices describing the linear quantum model .
this notion of physical realizability has been further investigated in @xcite where the authors prove the equivalence between the algebraic conditions for physical realizability obtained in @xcite and a frequency domain condition that an associated linear system is @xmath0-unitary .
in fact , @xcite extends the frequency domain physical realizability results of @xcite , @xcite and @xcite which apply to linear quantum systems described purely in terms of annihilation operators .
more explicitly , in @xcite , @xcite and @xcite , the physical realizability conditions developed for annihilation - operator linear quantum systems have been related to the lossless property of linear systems and @xcite generalizes this result to relate the corresponding physical realizability conditions of annihilation and creation operators linear quantum systems to the @xmath0-unitary property . in this paper , we restrict attention to a class of nonlinear quantum systems which is a generalization of the annihilation operator only linear quantum systems considered in @xcite , @xcite and @xcite . then , we relate the resulting physical realizability conditions to the lossless property of nonlinear dissipative systems having a specific type of storage function .
we consider an open quantum system g with physical variable space @xmath1 consisting of operators defined on an underlying hilbert space @xmath2 .
the self - energy of this system is described by a hamiltonian @xmath3 .
the system is driven by @xmath4 field channels given by the quantum stochastic process @xmath5
@xmath6 these describe the annihilation of photons in the field channels and are operators on a hilbert space @xmath7 , with associated variable space @xmath8 . in that case
, @xmath7 is the hilbert space defining an indefinite number of quanta ( called a fock space @xcite ) , and @xmath8 is the space of operators over this space .
we assume the process @xmath5 is canonical , meaning that we have the following second order ito products : @xmath9 where @xmath10 is the operator adjoint of @xmath11 which is defined on the same fock space .
the system is coupled to the field through a scattering matrix @xmath12 and a coupling vector of operators @xmath13 given by @xmath14 where @xmath15 .
the notation @xmath16 is used to indicate an open quantum system specified by the parameters @xmath17 and @xmath18 where @xmath18 represents the hamiltonian of the system . for @xmath12 ,
the schrodinger equation is given by @xmath19 with the initial condition @xmath20 .
equation ( [ schrodinger ] ) determines the unitary motion of the system in accordance of the fundamental laws of quantum mechanics .
note that the notation @xmath21 refers to the hilbert space adjoint . given a system annihilation operator @xmath22 , its heisenberg evolution is defined by @xmath23 and satisfies @xmath24\right)\\\nonumber & & + \left[l(t)^\dagger , a_l(t)\right]dw(t)\end{aligned}\ ] ] where the notation @xmath25=ab - ba$ ] denotes the commutator of two operators @xmath26 and @xmath27 . in equation ( [ annihi_eq ] )
, all operators evolve unitarily and the notation @xmath28 refers to @xmath29 + \frac{1}{2 } \left[l^\dagger , a_l\right]l.\ ] ] in this paper , we restrict attention to annihilation only coupling operators ; i.e , the coupling operator @xmath30 depends only on the annihilation operator @xmath31 and not the creation operator @xmath32 .
therefore , @xmath30 satisfies the following commutation property : @xmath33=0 $ ] and then @xmath34l.\ ] ] the generator of the system @xmath35 is given by @xmath36+\mathcal{l}_l(a_l).\ ] ] for the case of having @xmath37 annihilation operators @xmath38 , we define @xmath39 @xmath40l\ ] ] and @xmath41+\mathcal{l}_{l_i}(a).\ ] ] therefore , we can write @xmath42\right)\\\nonumber & & + \left[l(t)^\dagger , a_i(t)\right]dw(t)\end{aligned}\ ] ] and @xmath43\right)\\\nonumber & & + \left[l(t)^\dagger , a(t)\right]dw(t)\end{aligned}\ ] ] where @xmath44 and @xmath45= \left ( { \begin{array}{*{20}{c } } { \begin{array}{*{20}{c } } { { \left[a_1,h\right]}}\\ \vdots \\ { { \left[a_n , h\right ] } } \end{array } } \end{array } } \right).\ ] ] let @xmath46}}\\ \vdots \\ { { \left[a_n , h\right ] } } \end{array } } \end{array } } \right)\\\nonumber & = & \mathcal{l}_{l}(a)-i\left[a , h\right]\\ & = & \mathcal{g}_g(a)\end{aligned}\ ] ] and @xmath47}}\\ \vdots \\ { { \left[l^\dagger , a_n\right ] } } \end{array } } \end{array}}\right)\\&= & \left[l^\dagger , a\right].\end{aligned}\ ] ] hence , @xmath48 note that for the case of matrices , the notations @xmath49 and @xmath21 refer respectively to the complex conjugate and the complex conjugate transpose of the matrix in question . @xmath50 on the other hand , the components of the output fields are defined by @xmath51 and satisfy the nonlinear quantum stochastic differential equations @xmath52 @xmath53 where @xmath54 , @xmath55 , @xmath56 and @xmath57 .
hence , the system @xmath35 can be described by the following nonlinear quantum stochastic differential equations @xmath58 where @xmath59 $ ] , @xmath60 $ ] , @xmath61 $ ] , @xmath62 $ ] , @xmath63 $ ] and @xmath64 $ ] .
[ def4 ] the class of nonlinear quantum system we consider in this paper are nonlinear quantum systems that can be represented by the qsdes ( [ sys_a ] ) such that the matrices @xmath65 and @xmath66 satisfy @xmath67=0 $ ] , @xmath68=0 $ ] , @xmath69=-\left[a , a(a , a^\dagger)^t\right]$ ] and @xmath70 for @xmath71 and @xmath72 with @xmath73 , @xmath74 , @xmath75 , @xmath76 and @xmath77 integers .
moreover , the matrices @xmath78 , @xmath79 and @xmath80 satisfy the following property : @xmath81 - \frac{1}{{{2{\bar n}}}}\left [ { { { \bar a}^\dag } { { \bar \theta } ^ { - 1}}\bar a(a,{a^\dag } ) , \bar a } \right]=\\ & & \qquad \bar a(a,{a^\dag } ) - \frac{1}{2}\bar b(a,{a^\dag } ) \bar c(a , a^\dagger)\end{aligned}\ ] ] where @xmath82 , @xmath83 $ ] and @xmath84 $ ] . for more information , on @xmath85 , refer to section iii below . in the sequel
, we assume that @xmath86 admits the following decomposition : @xmath87 where @xmath88 is the noise part of @xmath89 and @xmath90 is an adapted process ( see @xcite and @xcite ) .
the noise @xmath88 is a vector of quantum weiner processes with ito table @xmath91 ( see @xcite ) where @xmath92 is a non - negative definite hermitian matrix . here
, the notation @xmath21 represents the adjoint transpose of a vector of operators . also , we assume the following commutation relations hold for the noise components : @xmath93 \triangleq d\tilde w ( t)d\tilde w(t)^{\dagger } - ( d\tilde w(t)^
* d\tilde w(t)^t)^t = t_w dt.\ ] ] here @xmath94 is a hermitian matrix and the notation @xmath95 denotes the transpose of a vector or matrix of operators .
the noise processes can be represented as operators on an appropriate fock space ; for more details , see @xcite .
the process @xmath90 represents variables of other systems which may be passed to the system ( [ sys_a ] ) via an interaction .
therefore , it is required that @xmath96 be an operator on a hilbert space distinct from that of @xmath97 and the noise processes .
we also assume that @xmath90 commutes with @xmath98 for all @xmath99 .
moreover , since @xmath90 is an adapted process , we note that @xmath90 also commutes with @xmath100 for all @xmath101 .
we have also that @xmath102=0 $ ] .
for the nonlinear quantum system ( [ sys_a ] ) , the initial system variables @xmath103 consist of operators satisfying the commutation relations @xmath104 & = & \theta _ { jk } , \qquad j , k = 1 , \ldots , n.\\ \left [ { \bar a_j ( 0),\bar a_k^{*}(0 ) } \right ] & = & \bar \theta _ { jk } , \qquad j , k = 1 , \ldots , 2n.\end{aligned}\ ] ] here , the commutator is defined by @xmath105 \triangleq a_ja_k^ { * } - a_{k}^ * a_j = \theta_{jk}$ ] where @xmath85 is a complex matrix with elements @xmath106 . with @xmath107 ,
the relations ( [ relations ] ) can be written as @xmath108 & \triangleq & aa^\dagger - ( a^*a^t ) ^t= \theta\\ \left[\bar a , \bar a^\dagger\right ] & \triangleq & \left [ { \begin{array}{*{20}{c } } \theta & 0\\ 0&{{\theta ^ * } } \end{array } } \right]=\bar \theta.\end{aligned}\ ] ] the following theorem provides an algebraic characterization of when the nonlinear quantum system ( [ sys_a ] ) preserves the commutation relations ( [ relations ] ) as time evolves .
[ thm1 ] for the nonlinear quantum system ( [ sys_a ] ) , we have that @xmath109=\bar \theta_{pq}$ ] implies @xmath110=\bar \theta_{pq}$ ] for all @xmath101 with @xmath111 if and only if @xmath112+\left[\bar a , \bar a(\bar a,\bar a^\dagger)^\dagger \right]\\\nonumber & & \quad + \bar b(\bar a , \bar a^\dagger)t _ { \bar w}\bar b(\bar a,\bar a^\dagger)^\dagger = 0;\\\nonumber & & \left [ \bar b(\bar a,\bar a^\dagger),\bar a^\dagger\right ] = 0 \qquad \mbox{and}\\ & & \left [ \bar a , \bar b(\bar a,\bar a^\dagger)^\dagger \right]=0\end{aligned}\ ] ] with @xmath113 $ ] .
in this section , we consider conditions under which a nonlinear quantum system of the form ( [ sys_a ] ) corresponds to a nonlinear open quantum harmonic oscillator .
such a system is said to be physically realizable .
the class of nonlinear open quantum harmonic oscillators under consideration are defined by the nonlinear coupling operator @xmath114 and a nonlinear hamiltonian @xmath115 . to derive a nonlinear system of the form ( [ sys_a ] ) from a nonlinear open quantum harmonic oscillator defined by @xmath116 and @xmath117 , we proceed as in section ii to get the following definition .
[ def1 ] a nonlinear quantum system of the form ( [ sys_a ] ) is said to be physically realizable if it is a representation of a nonlinear open quantum harmonic oscillator defined by a nonlinear coupling operator @xmath116 and a nonlinear hamiltonian @xmath117 , i.e. , if there exist @xmath117 and @xmath116 such that the matrices @xmath118 and @xmath119 satisfy the following equations @xmath120\bar
i \bar l + i\left[\bar h , \bar a\right ] ; \\\nonumber \bar b(a , a^\dagger)&=&\left[\bar l^\dagger , \bar a\right]\bar i;\\\nonumber \bar c(a , a^\dagger)&=&\bar l;\\
\bar d(a , a^\dagger)&=&i;\end{aligned}\ ] ] where @xmath121.$ ] the following theorem provides necessary and sufficient conditions for a nonlinear quantum system of the form ( [ sys_a ] ) to be physically realizable . [ thm2 ]
a nonlinear quantum system of the form ( [ sys_a ] ) is physically realizable with @xmath122 $ ] if and only if @xmath123+\left[\bar a , \bar a ( a , a^\dagger)^\dagger \right]\\\nonumber & & \quad + \bar b ( a , a^\dagger)\bar i \bar b ( a , a^\dagger)^\dagger = 0;\\\nonumber & & \left [ \bar b ( a , a^\dagger),\bar a^\dagger \right ] = 0;\\\nonumber & & \left [ \bar a , \bar b ( a , a^\dagger)^\dagger \right]=0;\\\nonumber & & \bar b ( a , a^\dagger)= \left[\bar c(a , a^\dagger)^\dagger,\bar a \right]\bar i\qquad \mbox{and}\\ & & \bar d ( a , a^\dagger)=i.\end{aligned}\ ] ] in this case , the corresponding nonlinear self - adjoint hamiltonian @xmath117 is given by @xmath124 where @xmath125 is as defined in definition [ def4 ] and the corresponding coupling operator is @xmath126 . _ * note : * _ note that the physical realizability conditions as illustrated by conditions ( [ phys_cond1 ] ) impose the restriction of requiring @xmath80 to be a vector of first order polynomials of the annihilation operator @xmath127 ; i.e , the index @xmath128 in definition [ def4 ] is only @xmath129 and @xmath130 for @xmath72 .
consequently , this in turn imposes the restriction of requiring the matrix @xmath79 to be a constant matrix .
in this section , we consider a class of nonlinear quantum systems of the form ( [ sys_a ] ) and we give a quantum version of the lossless property for this class .
[ def2 ] given an operator valued quadratic form @xmath131 where @xmath132 , @xmath133 and @xmath134 are constant matrices with @xmath135 and @xmath136 hermitian , we say that the system ( [ sys_a ] ) is dissipative with supply rate @xmath137 if there exists a storage function @xmath138 which is polynomial operator valued mapping with @xmath139 for all @xmath140 such that @xmath141 and @xmath142 for all @xmath101 and all operator valued @xmath143 .
[ def3 ] the nonlinear quantum system of the form ( [ sys_a ] ) is said to be lossless if it is dissipative with @xmath144 , @xmath145 and @xmath146 and @xmath147 whenever @xmath148 for all operator valued @xmath143 and all @xmath101 .
[ thm3 ] a necessary and sufficient condition for the nonlinear quantum system ( [ sys_a ] ) to be lossless with respect to the supply rate ( [ rate ] ) is that there exists a storage function @xmath149 with @xmath139 for all @xmath140 such that @xmath141 and @xmath150
in this section , we show that the nonlinear quantum system of the form ( [ sys_a ] ) is physically realizable if and only if it is lossless with a specific storage function .
[ thm4 ] the nonlinear quantum system ( [ sys_a ] ) is physically realizable if and only if it is lossless with a differentiable storage function @xmath149 satisfying @xmath151&= & \left [ { \begin{array}{*{20}{c } } 0&{2i}\\ { -2i}&0 \end{array } } \right].\end{aligned}\ ] ]
we consider a nonlinear quantum system described by @xmath152;\\\nonumber b(a , a^\dagger)&= & \left [ { \begin{array}{*{20}{c } } { - \sqrt { 2{k_{{a_1 } } } } } & 0\\ 0 & { - \sqrt { 2{k_{{a_2 } } } } } \end{array } } \right];\\\nonumber c(a , a^\dagger)&= & \left [ { \begin{array}{*{20}{c } } { \sqrt { 2{k_{{a_1 } } } } } a_1\\ { \sqrt { 2{k_{{a_2 } } } } } a_2 \end{array } } \right];\\\nonumber d(a , a^\dagger)&= & i.\end{aligned}\ ] ] where @xmath127 is the annihilation operator , @xmath153 is the creation operator and @xmath154 and @xmath155 are constants . hence , @xmath156;\\\nonumber \bar b(a , a^\dagger)&= & \left [ { \begin{array}{*{20}{c } } { - \sqrt { 2{k_{{a_1 } } } } } & 0&0&0\\ 0 & { - \sqrt { 2{k_{{a_2 } } } } } & 0 & 0\\ 0 & 0 & { -\sqrt { 2{k_{{a_1 } } } } } & 0\\ 0&0 & 0 & { -\sqrt { 2{k_{{a_2 } } } } } \end{array } } \right];\\\nonumber \bar c(a , a^\dagger)&= & \left [ { \begin{array}{*{20}{c } } { \sqrt { 2{k_{{a_1 } } } } } a_1\\ { \sqrt { 2{k_{{a_2 } } } } } a_2\\ { \sqrt { 2{k_{{a_1 } } } } } a_1^*\\ { \sqrt { 2{k_{{a_2 } } } } } a_2^ * \end{array } } \right];\\\nonumber \bar d(a , a^\dagger)&= & i.\end{aligned}\ ] ] the matrices @xmath78 , @xmath79 and @xmath157 satisfy the properties of the class of nonlinear quantum systems under consideration as given in definition [ def4 ] .
in fact , @xmath158=0 $ ] , @xmath68=0 $ ] and @xmath69=-\left[a , a(a , a^\dagger)^t\right]= \left [ { \begin{array}{*{20}{c } } { - 2a_2 ^ 2}&0\\ 0&{2a_1 ^ 2 } \end{array } } \right]$ ] .
moreover , @xmath78 and @xmath157 satisfy conditions ( [ prop_matrices ] ) in definition [ def4 ] . in this case ,
@xmath159 .
also , condition ( [ prop_matrices_2 ] ) in definition [ def4 ] is satisfied . in fact , @xmath160= \left [ { \begin{array}{*{20}{c } } { a_1^*a_2 ^ 2}\\ { - a_2^*a_1 ^ 2}\\ { a_2^{2*}{a_1}}\\ { - a_1^{2*}{a_2 } } \end{array } } \right]$ ] and @xmath161=\left [ { \begin{array}{*{20}{c } } { -a_1^*a_2 ^ 2}\\ { a_2^*a_1 ^ 2}\\ { -a_2^{2*}{a_1}}\\ { a_1^{2*}{a_2 } } \end{array } } \right]$ ] . on the other hand , @xmath162 $ ] .
hence , @xmath163 - \frac{1}{{{2{\bar n}}}}\left [ { { { \bar a}^\dag } { { \bar \theta } ^ { - 1}}\bar a(a,{a^\dag } ) , \bar a } \right]=\\\nonumber & & \qquad \bar a(a,{a^\dag } ) - \frac{1}{2}\bar b(a,{a^\dag } ) \bar c(a , a^\dagger).\end{aligned}\ ] ] in addition , the system under consideration is physically realizable .
in fact , @xmath164\bar i= \left [ { \begin{array}{*{20}{c } } { - \sqrt { 2{k_{{a_1 } } } } } & 0&0&0\\ 0 & { - \sqrt { 2{k_{{a_2 } } } } } & 0&0\\ 0&0&{-\sqrt { 2{k_{{a_1 } } } } } & 0\\ 0&0&0&{-\sqrt { 2{k_{{a_2 } } } } } \end{array } } \right]$ ] . on the other hand , @xmath165 = 0 $ ] ; @xmath166=0 $ ] and @xmath167 . also , @xmath168 = \left [ { \begin{array}{*{20}{c } } { - { k_{{a_1}}}}&{4a_1^*{a_2 } } & { - 2a_2 ^ 2}&0\\ { - 4a_2^*{a_1 } } & { - { k_{{a_2}}}}&0&{2a_1 ^ 2}\\ { 2a_2^{2*}}&0&{{k_{{a_1 } } } } & { - 4{a_1}a_2^*}\\ 0 & { - 2a_1^{2*}}&{4{a_2}a_1^*}&{{k_{{a_2 } } } } \end{array } } \right]$ ] , @xmath169 = \left [ { \begin{array}{*{20}{c } } { - { k_{{a_1 } } } } & { - 4a_1^*{a_2}}&{2a_2 ^ 2}&0\\ { 4a_2^*{a_1 } } & { - { k_{{a_2}}}}&0 & { - 2a_1 ^ 2}\\ { - 2a_2^{2*}}&0&{{k_{{a_1}}}}&{4{a_1}a_2^*}\\ 0&{2a_1^{2 * } } & { - 4{a_2}a_1^*}&{{k_{{a_2 } } } } \end{array } } \right]$ ] and @xmath170 $ ] . hence , @xmath171+\left[\bar a , \bar a ( a , a^\dagger)^\dagger \right]+ \bar b ( a , a^\dagger)\bar i \bar b ( a , a^\dagger)^\dagger = 0.$ ] in this case , the corresponding nonlinear self - adjoint hamiltonian @xmath117 is given by @xmath172 and the corresponding coupling operator is @xmath126 . moreover , the system is lossless .
in fact , let @xmath173 .
hence , @xmath174 $ ] .
hence , @xmath175=\left [ { \begin{array}{*{20}{c } } 0&{2i}\\ { -2i}&0 \end{array } } \right]$ ] . also , @xmath176 .
moreover , @xmath177 and @xmath178 .
hence , the system is lossless using theorem [ thm3 ] .
in this paper , the issue of physical realizability for a class of nonlinear quantum systems is developed and was found to be connected to the lossless properties of the system .
this provides a connection between the quantum mechanical and nonlinear system theory properties of the system . c. w. gardiner and m. j. collett , `` input and output in damped quantum systems : quantum stochastic differential equations and the master equation , '' _ physical review a _ , vol .
31 , no . 6 , pp .
37613774 , 1985 .
a. j. shaiju and i. r. petersen , `` on the physical realizability of general linear quantum stochastic differential equations with complex coefficients , '' _ proceedings of the joint 48th ieee conference on decision and control and the 28th chinese control conference , shanghai , p.r .
china , december 16 - 18 , 2009 _ , pp . 14221427 , 2009 .
s .z .s .
hassen and i. r. petersen and e. h. huntington and m. heurs and m. r. james , `` @xmath181 control of an optical squeezer , '' _ proceedings of the american control conference acc 2010 , baltimore , usa _
, pp.2730 - 2735 , 2010 . | in this paper , the physical realizability condition for a specific class of nonlinear quantum systems is related to the lossless property of nonlinear dissipative systems having a specific storage function
. [ section ] [ section ] |
the ultra - relativistic heavy ion collision starts out as a state of high density nuclear matter called the quark gluon plasma(qgp ) and expands rapidly to freeze - out . during the freeze - out phase quarks and gluons
form a system of strongly interacting hadrons .
these hadrons continue to expand in a thermal manner until no further scattering is possible because the system becomes to dilute .
however this transition from quarks and gluons(partons ) into hadrons is not a smooth affair .
the expansion is very rapid and some faster or hard scattered partons fragment directly into hadron through a minijet@xcite process .
thus we have thermal and minijet hadrons present in the last scattering of the hadrons .
the dipion cocktail part 1 considered this mixture of sources and applied it to the dipion mass spectrum of the heavy ion fireball formed in au - au collisions at @xmath1=200 .
part 1 showed that both thermal or soft production of hadrons and the minijet fragmented hadrons can be described through a set of unified formal equations .
part 2(this paper ) applies this formalism to the @xmath0 dependence of dipions for au - au collisions at @xmath4 200 gev and 40% to 80% centrality .
the paper is organized in the following manner : sec .
1 introduction .
sec . 2 review of the two component model which we use to fit the dipion data within a set of @xmath0 ranges .
3 discussion of the relationship between the @xmath2 parameter and the mass and widths of resonances .
4 we present a fit to 19 @xmath0 ranges for au - au collisions at @xmath4 200 gev and 40% to 80% centrality .
sec . 5 summary and discussion .
in this section we will alter equation 6 of part 1 so it can use breit - wigner parameters ( mass , width ) instead of phase shifts .
we will also need to modify the re - scattering part of the equation in order to have the correct threshold behavior we have introduced in part 1 for the minijet partial waves .
the phase shift can be written for the @xmath5 wave as @xmath6 where @xmath7 is the mass of the resonance in the @xmath8wave and @xmath9 is its total width .
@xmath10 with @xmath11 the total width at resonance , @xmath12 is the blatt - weisskopf - barrier factor@xcite for the @xmath8 of the resonance , @xmath13 is the @xmath14 center mass momentum , @xmath15 is @xmath13 at resonance , @xmath7 is the mass of the resonance , and @xmath16 is center mass momentum related to the size(1.0 fm is used @xmath16 = .200 gev / c ) .
using equation 1 we rewrite equation 6 of part 1 as @xmath17 the @xmath18 is the thermal production term and is constant except for the boltzmann weight(see equation 13 in part 1 ) .
the expected threshold behavior @xmath19 comes from the @xmath20 term .
since there is @xmath21 one of the @xmath19 is killed off by dividing by @xmath22 . in figure 6 of part 1
we see we have put into our minijet @xmath23 the correct threshold @xmath19 so we need to kill off the @xmath19 of the other @xmath20 term .
therefore the above equation for our minijet @xmath23 we will use @xmath24 rewriting equation 6 of part 1 for each partial wave with breit - wigner parameters the first term becomes @xmath25 while the second term @xmath26
@xmath27 where @xmath28 and @xmath29,@xmath30,@xmath31 , and @xmath32 . s , p , d and f comes from subsection 5.2 of part 1 .
equation 6 of part 1 has an important factor the coefficient @xmath2 .
this coefficient is related to the real part of the @xmath33 re - scattering loop and is given by equation 9 .
when the pions re - scatter or interact at a close distance or a point the real part of the loop @xmath2 has its maximum value of @xmath34 .
while if the pions re - scatter or interact at a distance determined by the diffractive limit the value of @xmath2 is zero .
the @xmath2 which is the real part of the re - scattering factor has a simple form given by @xmath35 where @xmath36 is the radius of re - scattering in fm s and @xmath37 is 1.0 fm or the limiting range of the strong interaction ranging to @xmath36 = 0.0 for point like interactions
. when @xmath33 pairs interact at the diffractive limit their phase shift should be the same as the phase shift of the vacuum .
the same statement is true for @xmath33 interacting at a point since asymptotic freedom demands that the strong interaction should have no effect .
however values of @xmath2 in between zero and @xmath34 represent a confined volume where strongly interacting gluons , quarks and virtual mesons may influence the phase shift of the @xmath33 system .
phase shifts under a breit - wigner parameters assumption depend on the mass and width of the resonance parameter . in the next section we use
fits to data to determine the relationship between breit - wigner parameters and the value of @xmath2 .
we have fitted 19 dipion @xmath0 ranges(see table i ) using equation 7 above for the star au - au collisions at @xmath4 200 gev 40% to 80% centrality data .
we included minijets up to @xmath8 = 3 and resonances @xmath39 @xmath8 = 0 , @xmath40 @xmath8 = 1 , and @xmath41 @xmath8 = 2 . using the arguments of sec .
3 of part 1 , we added the @xmath42 as a direct thermal term ( @xmath43 ) and only the @xmath39 interfered with @xmath8 = 0 minijet background .
two other thermal terms are present in the cocktail , the @xmath44 and the @xmath45 .
all the thermal terms have an exponential behavior with dipion @xmath0 .
the spectrum of the minijet partial waves is obtained from pythia@xcite(see sec .
5.2 of part 1 ) .
we let the data determine which minijet partial wave to add .
we find only swave minijet background is important until @xmath0 equal to 1.1 gev / c . above 1.5 gev
/ c all four minijet partial waves are used up to fwave .
it should be noted dwave and fwave are small effects .
we used pdg@xcite for the @xmath41 mass = 1.275 gev and width = .185 gev .
the @xmath42 was fitted obtaining mass = 0.9727 @xmath46 .0039 gev and width = 0.04512 @xmath46 0.01128 gev .
the @xmath39 mass and width used was fixed because it was ill determined .
the mass used was mass = 1.011 gev and width = 1.015 gev .
the @xmath3 mass and width is explained below .
finally the threshold effective mass region .280 gev to .430 gev is dominated by the swave and receives contributions from minijet fragmentation , @xmath33 swave phase shift , @xmath47 decay , hbt adding to the like sign @xmath33 distribution that has been subtracted away from the unlike sign @xmath33 and the coulomb correction between the charged pions .
the minijet fragmentation is the least known of the effects since we relied on pythia , however there are large uncertainty in all the other effects .
so for these fits we let the minijet fragmentation be free to fit the data and let the breit - wigner parameters for the @xmath39 determine the swave phase shifts plus leaving out all other effects . for the @xmath2 parameter in @xmath0
bins up to 1.1 gev / c the minijet swave interference is the determining factor . above 1.1 gev / c the pwave interference becomes most important .
the values of @xmath2 which gives a reasonable fit are shown in table ii .
we have determined that the @xmath39 pole or breit - wigner parameters is so far away from the real axis thus it is too short lived to be influenced by hadronic interactions .
the @xmath3 phase shift being of a life time comparable to hadronic interaction taking place becomes most sensitive .
we have found as a function of @xmath2 the best of @xmath3 width is always 0.147 gev with an error of @xmath46 .007 gev .
the mass however decreases as @xmath2 grows , reaching a minimum of 0.738 gev at an @xmath2 of 0.907 .
this is a mass shift of 37 mev .
an @xmath2 of 0.504 is the smallest @xmath2 we find in our fits .
a mass of 0.775 gev is the best fit when the value of @xmath2 is at 0.504 . using equation 9 in table ii we determine the radius of @xmath33 re - scattering for each @xmath0 bin .
the value of @xmath34 used in table ii is equal to 2.0 as determined in appendix b of part 1 .
table ii shows an interesting density effect around dipion @xmath0 of 0.6 to 1.0 gev / c . if one consider that @xmath0 maybe related to fireball size through the idea of hubble flow , then pions with a @xmath0 of around 0.4 gev / c maybe coming from a less dense region in the central part of the fireball .
this could be a density wave effect . * table i. the @xmath0 ranges bins we fit . *
@xmath0 bin number & lower edge(gev / c ) & upper edge(gev / c ) + 1 & 0.2 & 0.4 + 2 & 0.4 & 0.6 + 3 & 0.6 & 0.8 + 4 & 0.8 & 1.0 + 5 & 1.0 & 1.2 + 6 & 1.2 & 1.4 + 7 & 1.4 & 1.6 + 8 & 1.6 & 1.8 + 9 & 1.8 & 2.0 + 10 & 2.0 & 2.2 + 11 & 2.2 & 2.4 + 12 & 2.4 & 2.6 + 13 & 2.6 & 2.8 + 14 & 2.8 & 3.0 + 15 & 3.0 & 3.2 + 16 & 3.2 & 3.4 + 17 & 3.4 & 3.6 + 18 & 3.6 & 3.8 + 19 & 3.8 & 4.0 + * table ii .
the @xmath2 value , @xmath3 mass value and radius in each @xmath0 range . *
@xmath0(gev / c ) & @xmath2 & @xmath3 mass(gev ) & radius(f ) + 0.3 & 0.907 @xmath46 .028 & 0.738 @xmath46 .004 & .739 @xmath46 .010 + 0.5 & 0.806 @xmath46 .025 & 0.748 @xmath46 .003 & .773 @xmath46 .008 + 0.7 & 0.706 @xmath46 .022 & 0.755 @xmath46 .002 & .804 @xmath46 .007 + 0.9 & 0.706 @xmath46 .022 & 0.755 @xmath46 .002 & .804 @xmath46 .007 + 1.1 & 0.806 @xmath46 .025 & 0.748 @xmath46 .003 & .773 @xmath46 .008 + 1.3 & 0.806 @xmath46 .025 & 0.748 @xmath46 .003 & .773 @xmath46 .008 + 1.5 & 0.806 @xmath46 .025 & 0.748 @xmath46 .003 & .773 @xmath46 .008 + 1.7 & 0.806 @xmath46 .025 & 0.748 @xmath46 .003 & .773 @xmath46 .008 + 1.9 & 0.806 @xmath46 .025 & 0.748 @xmath46 .003 & .773 @xmath46 .008 + 2.1 & 0.605 @xmath46 .019 & 0.759 @xmath46 .002 & .835 @xmath46 .006 + 2.3 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 2.5 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 2.7 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 2.9 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 3.1 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 3.3 & 0.504
@xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 3.5 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 3.7 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + 3.9 & 0.504 @xmath46 .016 & 0.775 @xmath46 .001 & .865 @xmath46 .004 + the 19 dipion spectrum for the @xmath0 bins are shown in figure 1 through figure 19 .
the direct cross sectional yield of the thermally produced states into @xmath48 @xmath49 is shown in figure 20 .
the states are @xmath50 , @xmath3 , @xmath42 and @xmath51 and the yields come from the exponential fits to the direct thermal component of equation 7 .
the dipion cocktail part 2 applies a formalism derived in the dipion cocktail part 1 which considered a mixture of sources present in the dipion mass spectrum of the heavy ion fireball .
part 1 showed that both thermal or soft production of hadrons and the minijet fragmented hadrons can be described through a set of unified formal equations .
part 2(this paper ) applies this part 1 formalism to the @xmath0 dependence of dipions for au - au collisions at @xmath4 200 gev and 40% to 80% centrality .
part 1 started with the basic definition of elastic @xmath33 scattering .
next showed how re - scattering of pions depends on the unitary condition that interactions present in the phase shift of an orbital state must interact all the time .
the process of parton fragmentation into dipion states through unitarity leads to a equation of production and re - scattering in a given orbital quantum number .
this equation(equation 7 ) has two components in each orbital state : one being the thermal production of resonances in a dipion orbital state , the other is the re - scattering of dipions coming from parton or minijet fragmentation into the dipion orbital state which do not come directly from the resonance .
unitarity requires that there most be re - scatter through resonance phase shifts which we defined through breit - wigner parameters ( mass , width ) .
we have fitted 19 dipion @xmath0 ranges(see table i ) using equation 7 .
we included minijets up to @xmath8 = 3 and resonances @xmath39 @xmath8 = 0 , @xmath40 @xmath8 = 1 , and @xmath41 @xmath8 = 2 . using the arguments of sec .
3 of part 1 , we added the @xmath42 as a direct thermal term ( @xmath43 ) and only the @xmath39 interfered with @xmath8 = 0 minijet background .
two other thermal terms are present in the cocktail , the @xmath44 and the @xmath45 .
all the thermal terms have an exponential behavior with dipion @xmath0 and are shown figure 20 .
the spectrum of the minijet partial waves is obtained from pythia@xcite(see sec .
5.2 of part 1 ) .
we let the data determine which minijet partial wave to add .
we find only swave minijet background is important until @xmath0 equal to 1.1 gev / c . above
1.5 gev / c all four minijet partial waves are used up to fwave .
it should be noted dwave and fwave are small effects .
we used pdg@xcite for the @xmath41 mass = 1.275 gev and width = .185 gev .
the @xmath42 was fitted obtaining mass = 0.9727 @xmath46 .0039 gev and width = 0.04512 @xmath46 0.01128 gev .
the @xmath39 mass and width used was fixed because it was ill determined .
the mass used was mass = 1.011 gev and width = 1.015 gev . for the @xmath2 parameter in @xmath0
bins up to 1.1 gev / c the minijet swave interference is the determining factor . above 1.1 gev / c the pwave interference becomes most important .
the values of @xmath2 which gives a reasonable fit are shown in table ii .
we have determined that the @xmath39 pole or breit - wigner parameters is so far away from the real axis thus it is too short lived to be influenced by hadronic interactions .
the @xmath3 phase shift being of a life time comparable to hadronic interaction taking place becomes most sensitive .
we have found as a function of @xmath2 the best of @xmath3 width is always 0.147 gev with an error of @xmath46 .007 gev .
the mass however decreases as @xmath2 grows , reaching a minimum of 0.738 gev at an @xmath2 of 0.907 .
this is a mass shift of 37 mev .
an @xmath2 of 0.504 is the smallest @xmath2 we find in our fits .
a mass of 0.775 gev is the best fit when the value of @xmath2 is at 0.504 . using equation 9 in table
ii we determine the radius of @xmath33 re - scattering for each @xmath0 range .
table ii shows an interesting density effect around dipion @xmath0 of 0.6 to 1.0 gev / c .
if one consider that @xmath0 maybe related to fireball size through the idea of hubble flow , then pions with a @xmath0 of around 0.4 gev / c maybe coming from a less dense region in the central part of the fireball .
this could be a density wave effect .
this research was supported by the u.s .
department of energy under contract no .
de - ac02 - 98ch10886 .
the author thanks william love for the star analysis of the angular correlation data from run 4 . also for his assistance in the production of figures .
it is sad that he is gone .
99 t. trainor , phys .
c 80 ( 2009 ) 044901 .
f. von hippel and c. quigg , phys .
rev . 5 ( 1972 ) 624 .
t. sjostrand , m. van zijil , phys . rev .
d 36 ( 1987 ) 2019 .
j. beringer et al .
( particle data group ) , j. phys .
d86 ( 2012 ) 010001 . | in this paper we use methods developed in part 1 of the dipion cocktail , to fit the @xmath0 dependence of dipions for mid - central au - au collisions at @xmath1=200 gev . for the minijet fragmentation part we use pythia fragmentation as described in part 1 . for the thermal resonance production we use an exponential growth behavior .
the interference between the direct production of dipion pairs from non - resonance minijet fragmentation and re - scattering through resonance states gives a measure of the size of the re - scattering region .
this size is contain in the @xmath2 parameter of part1 .
we assumed a relationship between the @xmath2 parameter and the mass shift of the @xmath3 resonance .
the data used for the fits comes from the rhic collider as measured in the star experiment .
= 0.4 cm = 0.0 in = -0.0 in = 9.0 in = 6.5 in = 0.1 in + |
electromagnetic form factors are fundamental observables which characterize the composite nature of the nucleon .
several decades of elastic form factor experiments with electron beams , including recent high - precision measurements at jefferson lab and elsewhere , have provided considerable insight into the detailed structure of the nucleon . in the standard one - photon exchange ( born )
approximation , the electromagnetic current operator is parameterized in terms of two form factors , usually taken to be the dirac ( @xmath9 ) and pauli ( @xmath10 ) form factors , @xmath11 where @xmath12 is the momentum transfer to the nucleon , and @xmath13 is the nucleon mass .
the resulting cross section depends on two kinematic variables , conventionally taken to be @xmath14 ( or @xmath15 ) and either the scattering angle @xmath16 , or the virtual photon polarization @xmath17 . in terms of the sachs electric and magnetic form factors , defined as @xmath18
the reduced born cross section can be written @xmath19 the standard method which has been used to determine the electric and magnetic form factors , particularly those of the proton , has been the rosenbluth , or longitudinal - transverse ( lt ) , separation method . since the form factors in eq .
( [ eq : sigmar ] ) are functions of @xmath3 only , studying the cross section as a function of the polarization @xmath20 at fixed @xmath3 allows one to extract @xmath21 from the @xmath20-intercept , and the ratio @xmath22 from the slope in @xmath20 , where @xmath23 is the nucleon magnetic moment .
the results of the rosenbluth measurements for the proton have generally been consistent with @xmath24 for @xmath25 gev@xmath26 @xcite .
the `` super - rosenbluth '' experiment at jefferson lab @xcite , in which smaller systematic errors were achieved by detecting the recoiling proton rather than the electron , as in previous measurements , is also consistent with the earlier lt results .
an alternative method of extracting the ratio @xmath27 has been developed recently at jefferson lab @xcite , in which a polarized electron beam scatters from an unpolarized target , with measurement of the polarization of the recoiling proton . from the ratio of the transverse to longitudinal recoil polarizations
one finds @xmath28 where @xmath29 and @xmath30 are the initial and final electron energies , and @xmath2 ( @xmath1 ) is the polarization of the recoil proton transverse ( longitudinal ) to the proton momentum in the scattering plane .
the polarization transfer experiments yielded strikingly different results compared with the lt separation , with @xmath31 over the same range in @xmath3 @xcite . recall that in perturbative qcd one expects @xmath32 at large @xmath3 ( or equivalently @xmath33 ) @xcite , so that these results imply a strong violation of scaling behavior ( see also refs .
@xcite ) . the question of which experiments are correct has been debated over the past several years .
attempts to reconcile the different measurements have been made by several authors @xcite , who considered whether 2@xmath34 exchange effects , which form part of the radiative corrections ( rcs ) , and which are treated in an approximate manner in the standard rc calculations @xcite , could account for the observed discrepancy .
an explicit calculation @xcite of the two - photon exchange diagram , in which nucleon structure effects were for the first time fully incorporated , indeed showed that around half of the discrepancy could be removed just by the nucleon elastic intermediate states .
a partonic level calculation @xcite subsequently showed that the deep inelastic region can also contribute significantly to the box diagram . in this paper
we further develop the methodology introduced in ref .
@xcite , and apply it to systematically calculate the 2@xmath34 exchange effects in a number of electron
nucleon scattering observables .
we focus on the nucleon elastic intermediate states ; inelastic contributions are discussed elsewhere @xcite . in sec .
ii we examine the effects of 2@xmath34 exchange on the ratio of electric to magnetic form factors in unpolarized scattering .
in contrast to the earlier analysis @xcite , in which simple monopole form factors were utilized at the internal @xmath35 vertices , here we parameterize the vertices by realistic form factors , and study the model - dependence of effects on the ratio @xmath27 due to the choice of form factors .
we also compare the results with data on the ratio of @xmath6 to @xmath7 scattering cross sections , which is directly sensitive to 2@xmath34 exchange effects . in sec .
iii we examine the effects of 2@xmath34 exchange on the polarization transfer reaction , @xmath36 , for both longitudinally and transversely polarized recoil protons .
we also consider the case of proton polarization normal to the reaction plane , which depends on the imaginary part of the box diagram .
since this is absent in the born approximation , the normal polarization provides a clean signature of 2@xmath34 exchange effects , even though it does not directly address the @xmath37 discrepancy . following the discussion of the proton , in sec .
iv we consider 2@xmath34 exchange corrections to the form factors of the neutron , both for the lt separation and polarization transfer techniques . applying the same formalism to the case of the @xmath8he nucleus , in sec .
v we compute the elastic contribution from the box diagram to the ratio of charge to magnetic form factors of @xmath8he . in sec .
vi we summarize our findings , and discuss future work .
in this section we outline the formalism used to calculate the 2@xmath34 exchange contribution to the unpolarized electron nucleon cross section , and examine the effect on the @xmath37 ratio extracted using lt separation . since there are in general three form factors that are needed to describe elastic @xmath38 scattering beyond 1@xmath34 exchange , we also evaluate the 2@xmath34 contributions to each of the form factors separately . in the final part of this section
, we examine the effect of the 2@xmath34 correction on the ratio of @xmath6 to @xmath39 elastic cross sections , which is directly sensitive to 2@xmath34 exchange effects . for the elastic scattering process
we define the momenta of the initial electron and nucleon as @xmath40 and @xmath41 , and of the final electron and nucleon as @xmath42 and @xmath43 , respectively , @xmath44 . the four - momentum transferred from the electron to the nucleon
is given by @xmath45 ( with @xmath46 ) , and the total electron and proton invariant mass squared is given by @xmath47 . in the born approximation , the amplitude can be written @xmath48 where @xmath49 is the electron charge , and @xmath50 is given by eq .
( [ eq : current ] ) . in terms of the amplitude @xmath51 ,
the corresponding differential born cross section is given by @xmath52 where @xmath53 is the reduced cross section given in eq .
( [ eq : sigmar ] ) , and the mott cross section for the scattering from a point particle is @xmath54 with @xmath29 and @xmath30 the initial and final electron energies , and @xmath55 the electromagnetic fine structure constant .
including radiative corrections to order @xmath56 , the elastic scattering cross section is modified as @xmath57 where @xmath58 includes one - loop virtual corrections ( vacuum polarization , electron and proton vertex , and two photon exchange corrections ) , as well as inelastic bremsstrahlung for real photon emission @xcite .
according to the lt separation technique , one extracts the ratio @xmath59 from the @xmath20 dependence of the cross section at fixed @xmath3 . because of the factor @xmath60 multiplying @xmath61 in eq .
( [ eq : sigmar ] ) , the cross section becomes dominated by @xmath21 with increasing @xmath3 , while the relative contribution of the @xmath61 term is suppressed . hence understanding the @xmath20 dependence of the radiative correction @xmath58 becomes increasingly important at high @xmath3 . as pointed out in ref .
@xcite , for example , a few percent change in the @xmath20 slope in @xmath62 can lead to a sizable effect on @xmath27 .
in contrast , as we discuss in sec .
iii below , the polarization transfer technique does not show the same sensitivity to the @xmath20 dependence of @xmath58 .
if we denote the amplitude for the one - loop virtual corrections by @xmath63 , then @xmath63 can be written as the sum of a `` factorizable '' term , proportional to the born amplitude @xmath51 , and a non - factorizable part @xmath64 , @xmath65 the ratio of the full cross section ( to order @xmath56 ) to the born can therefore be written as @xmath66 with @xmath58 given by @xmath67 in practice the factorizable terms parameterized by @xmath68 , which includes the electron vertex correction , vacuum polarization , and the infrared ( ir ) divergent parts of the nucleon vertex and two - photon exchange corrections , are found to be dominant .
furthermore , these terms are all essentially independent of hadronic structure .
however , as explained in ref .
@xcite , the contributions to the functions @xmath68 from the electron vertex , vacuum polarization , and proton vertex terms depend only on @xmath3 , and therefore have no relevance for the lt separation aside from an overall normalization factor .
hence , of the factorizable terms , only the ir divergent two - photon exchange contributes to the @xmath20 dependence of the virtual photon corrections .
the terms which do depend on hadronic structure are contained in @xmath64 , and arise from the finite nucleon vertex and two - photon exchange corrections .
for the case of the proton , the hadronic vertex correction was analyzed by maximon and tjon @xcite , and found to be @xmath69 for @xmath70 gev@xmath26 . since the proton vertex correction does not have a strong @xmath20 dependence , it will not affect the lt analysis , and can be safely neglected .
for the inelastic bremsstrahlung cross section , the amplitude for real photon emission can also be written in the form of eq .
( [ eq : m1 ] ) . in the soft photon approximation
the amplitude is completely factorizable .
a significant @xmath20 dependence arises due to the frame dependence of the angular distribution of the emitted photon .
these corrections , together with external bremsstrahlung , contain the main @xmath20 dependence of the radiative corrections , and are usually accounted for in the experimental analyses .
they are generally well understood , and in fact enter differently depending on whether the electron or proton are detected in the final state .
hence corrections beyond the standard @xmath71 radiative corrections which can lead to non - negligible @xmath20 dependence are confined to the 2@xmath34 exchange diagrams , illustrated in fig .
[ fig : diag ] , and are denoted by @xmath72 , which we will focus on in the following .
the 2@xmath34 exchange correction @xmath73 which we calculate is then essentially @xmath74 in principle the two - photon exchange amplitude @xmath72 includes all possible hadronic intermediate states in fig .
[ fig : diag ] . here
we consider only the elastic contribution to the full response function , and assume that the proton propagates as a dirac particle ( excited state contributions are considered in ref .
we also assume that the structure of the off - shell current operator is similar to that in eq .
( [ eq : current ] ) , and use phenomenological form factors at the @xmath35 vertices .
this is of course the source of the model dependence in the problem .
clearly this also creates a tautology , as the radiative corrections are also used to determine the experimental form factors .
however , because @xmath58 is a ratio , the model dependence cancels somewhat , provided the same phenomenological form factors are used for both @xmath51 and @xmath75 in eq .
( [ eq : delta_eff ] ) . the total 2@xmath34 exchange amplitude , including the box and crossed box diagrams in fig .
[ fig : diag ] , has the form @xmath76 where the numerators are the matrix elements @xmath77 and the denominators are products of propagators @xmath78 [ ( k - q)^2-\lambda^2 ] \nonumber\\ & & \times [ ( p_1-k)^2-m^2 ] [ ( p_2+k)^2-m^2]\ , \label{eq : dbox } \\ % d_{\rm x - box}(k ) & = & \left . d_{\rm box}(k)\right|_{p_1-k\rightarrow p_3+k}\ .
\label{eq : dxbox}\end{aligned}\ ] ] an infinitesimal photon mass @xmath79 has been introduced in the photon propagator to regulate the ir divergences .
the ir divergent part is of interest since it is the one usually included in the standard rc analyses .
the finite part , which is typically neglected , has been included in ref .
@xcite and found to have significant @xmath20 dependence .
the ir divergent part of the amplitude @xmath72 can be separated from the ir finite part by analyzing the structure of the photon propagators in the integrand of eq .
( [ eq : mbox ] ) .
the two poles , where the photons are soft , occur at @xmath80 and at @xmath81 .
the dominant ( ir divergent ) contribution to the integral ( [ eq : mbox ] ) comes from the poles , and one therefore typically makes the approximation @xmath82 with @xmath83 in this case the ir divergent contribution is proportional to the born amplitude , and the corresponding correction to the born cross section is independent of hadronic structure .
the remaining integrals over propagators can be done analytically . in the target rest frame
the total ir divergent two - photon exchange contribution to the cross section is found to be @xmath84 a result given by maximon and tjon @xcite .
the logarithmic ir singularity in @xmath79 is exactly cancelled by a corresponding term in the bremsstrahlung cross section involving the interference between real photon emission from the electron and from the nucleon . by contrast , in the standard treatment of mo and tsai ( mt ) @xcite a different approximation for the integrals over propagators is introduced . here , the ir divergent contribution to the cross section is @xmath85 where @xmath86 and @xmath87 .
the logarithmic dependence on @xmath79 is the same as eq .
( [ eq : deltair ] ) , however . as mentioned above , the full expression in eq .
( [ eq : mbox ] ) includes both finite and ir divergent terms , and form factors at the @xmath35 vertices . in ref .
@xcite the proton form factors @xmath9 and @xmath10 were expressed in terms of the sachs electric and magnetic form factors , @xmath88 with @xmath89 and @xmath90 both parameterized by a simple monopole form , @xmath91 , with the mass parameter @xmath92 related to the size of the proton . in the present analysis we generalize this approach by using more realistic form factors in the loop integration , consistent with the actual @xmath93 data .
the functions @xmath9 and @xmath10 are parameterized directly in terms of sums of monopoles , of the form @xmath94 where @xmath95 and @xmath96 are free parameters , and @xmath97 is determined from the normalization condition , @xmath98 .
the parameters @xmath95 and @xmath96 for the @xmath9 and @xmath10 form factors of the proton and neutron are given in table i. the normalization conditions are @xmath99 and @xmath100 for the proton , and @xmath101 and @xmath102 for the neutron , where @xmath103 and @xmath104 are the proton and neutron anomalous magnetic moments , respectively .
in practice we use the parameterization from ref .
@xcite , and fit the parameterized form factors a sum of three monopoles , except for @xmath105 , which is fitted with @xmath106 .
as discussed in the next section , the sensitivity of the results to the choice of form factor is relatively mild .
of course , one should note that the data to which the form factors are fitted were extracted under the assumption of 1@xmath34 exchange , so that in principle one should iterate the data extraction and fitting procedure for self - consistency .
however , within the accuracy of the data and of the 2@xmath34 calculation the effect of this will be small . to obtain the radiatively corrected cross section for unpolarized electron scattering the polarizations of the incoming and outgoing electrons and nucleons in eqs .
( [ eq : nbox ] ) and ( [ eq : nxbox ] ) need to be averaged and summed , respectively .
the resulting expression involves a product of traces in the dirac spaces of the electron and nucleon .
the trace algebra is tedious but straightforward .
it was carried out using the algebraic program form @xcite and verified independently using the program tracer @xcite .
we also used two independent mathematica packages ( feyncalc @xcite and formcalc @xcite ) to carry out the loop integrals .
the packages gave distinct but equivalent analytic expressions , which gave identical numerical results .
the loop integrals in eq .
( [ eq : mbox ] ) can be expressed in terms of four - point passarino - veltman functions @xcite , which have been calculated using spence function @xcite as implemented by veltman @xcite . in the actual calculations
we have used the ff program @xcite .
the results of the proton calculation are presented in the following section . in typical experimental analyses of electromagnetic form factor data
@xcite radiative corrections are implemented using the prescription of ref .
@xcite , including using eq .
( [ eq : deltairmt ] ) to approximate the 2@xmath34 contribution . to investigate the effect of our results on the data analyzed in this manner
, we will therefore compare the @xmath20 dependence of the full calculation with that of @xmath107 . to make the comparison meaningful
, we will consider the difference @xmath108 in which the ir divergences cancel , and which is independent of @xmath79 .
the results for the difference @xmath109 between the full calculation and the mt approximation are shown in fig . [
fig : delta ] for several values of @xmath3 from 1 to 6 gev@xmath26 .
the additional corrections are most significant at low @xmath20 , and essentially vanish at large @xmath20 . at the lower @xmath3 values
@xmath109 is approximately linear in @xmath20 , but significant deviations from linearity are observed with increasing @xmath3 , especially at smaller @xmath20 . in fig .
[ fig : ffcomp](a ) we illustrate the model dependence of the results by comparing the results in fig . [ fig : delta ] at @xmath110 and 6 gev@xmath26 with those obtained using a dipole form for the @xmath111 and @xmath112 form factors , with mass @xmath113 gev . at the lower , @xmath114 gev@xmath26 , value the model dependence is very weak , with essentially no change at all in the slope . for the larger value @xmath115 gev@xmath26 the differences are slightly larger , but the general trend of the correction remains unchanged .
we can conclude therefore that the model dependence of the calculation is quite modest .
also displayed is the correction at @xmath116 gev@xmath26 , which will be accessible in future experiments , showing significant deviations from linearity over the entire @xmath20 range .
the results are also relatively insensitive to the high-@xmath3 behavior of the @xmath37 ratio , as fig .
[ fig : ffcomp](b ) illustrates . here
the correction @xmath109 is shown at @xmath115 gev@xmath26 calculated using various form factor inputs , from parameterizations obtained by fitting only the lt - separated data @xcite , and those in which @xmath117 is constrained by the polarization transfer data @xcite .
the various curves are almost indistinguishable , and the dependence on the form factor inputs at lower @xmath3 is expected to be even weaker than that in fig .
[ fig : ffcomp](b ) .
the effect of the 2@xmath34 corrections on the cross sections can be seen in fig .
[ fig : sigr ] , where the reduced cross section @xmath53 , scaled by the square of the dipole form factor , @xmath118 is plotted as a function of @xmath20 for several fixed values of @xmath3 . in fig .
[ fig : sigr](a ) the results are compared with the slac data @xcite at @xmath119 , 4 , 5 and 6 gev@xmath26 , and with data from the `` super - rosenbluth '' experiment at jlab @xcite in fig .
[ fig : sigr ] ( b ) . in both cases the born level results ( dotted curves ) , which are obtained using the form factor parameterization of ref .
@xcite in which @xmath117 is fitted to the polarization transfer data @xcite , have slopes which are significantly shallower than the data . with the inclusion of the 2@xmath34 contribution ( solid curves ) , there is a clear increase of the slope , with some nonlinearity evident at small @xmath20 .
the corrected results are clearly in better agreement with the data , although do not reproduce the entire correction necessary to reconcile the rosenbluth and polarization transfer measurements .
to estimate the influence of these corrections on the electric to magnetic proton form factor ratio , the simplest approach is to examine how the @xmath20 slope changes with the inclusion of the 2@xmath34 exchange .
of course , such a simplified analysis can only be approximate since the @xmath20 dependence is only linear over limited regions of @xmath20 , with clear deviations from linearity at low @xmath20 and high @xmath3 .
in the actual data analyses one should apply the correction @xmath109 directly to the data , as in fig .
[ fig : sigr ] .
however , it is still instructive to obtain an estimate of the effect on @xmath27 by taking the slope over several ranges of @xmath20 . following ref .
@xcite , this can be done by fitting the correction @xmath120 to a linear function of @xmath20 , of the form @xmath121 , for each value of @xmath3 at which the ratio @xmath27 is measured .
the corrected reduced cross section in eq .
( [ eq : sigmar ] ) then becomes @xmath122 + \mu^2 \tau b / a \right ) \right]\ , \label{eq : sigrcor}\ ] ] where @xmath123 is the `` true '' form factor ratio , corrected for 2@xmath34 exchange effects , and @xmath124 is the `` effective '' ratio , contaminated by 2@xmath34 exchange .
note that in eqs .
( [ eq : sigrcor ] ) and ( [ eq : rtrue ] ) we have effectively linearized the quadratic term in @xmath20 by taking the average value of @xmath20 ( _ i.e. _ , @xmath125 ) over the @xmath20 range being fitted .
in contrast to ref .
@xcite , where the approximation @xmath126 was made and the quadratic term in @xmath20 neglected , the use of the full expression in eq .
( [ eq : rtrue ] ) leads to a small decrease in @xmath27 compared with the approximate form . the shift in @xmath27 is shown in fig . [ fig : gemp ] , together with the polarization transfer data .
we consider two ranges for @xmath20 : a large range @xmath127 , and a more restricted range @xmath128 .
the approximation of linear @xmath20 dependence of @xmath109 should be better for the latter , even though in practice experiments typically sample values of @xmath20 near its lower and upper bounds .
a proposed experiment at jefferson lab @xcite aims to test the linearity of the @xmath20 plot through a precision measurement of the unpolarized elastic cross section .
the effect of the 2@xmath34 exchange terms on @xmath27 is clearly significant . as observed in ref .
@xcite , the 2@xmath34 corrections have the proper sign and magnitude to resolve a large part of the discrepancy between the two experimental techniques .
in particular , the earlier results @xcite using simple monopole form factors found a shift similar to that in for the @xmath128 range in fig .
[ fig : gemp ] , which resolves around 1/2 of the discrepancy .
the nonlinearity at small @xmath20 makes the effective slope somewhat larger if the @xmath20 range is taken between 0.2 and 0.9 .
the magnitude of the effect in this case is sufficient to bring the lt and polarization transfer points almost to agreement , as indicated in fig .
[ fig : gemp ] . while the 2@xmath34 corrections clearly play a vital role in resolving most of the form factor discrepancy , it is instructive to understand the origin of the effect on @xmath27 with respect to contributions to the individual @xmath117 and @xmath129 form factors . in general the amplitude for elastic scattering of an electron from a proton , beyond the born approximation , can be described by three ( complex ) form factors , @xmath130 , @xmath131 and @xmath132 .
the generalized amplitude can be written as @xcite @xmath133 where @xmath134 and @xmath135 .
the functions @xmath136 ( both real and imaginary parts ) are in general functions of @xmath3 and @xmath20 . in the 1@xmath34 exchange limit
the @xmath137 functions approach the usual ( real ) dirac and pauli form factors , while the new form factor @xmath132 exists only at the 2@xmath34 level and beyond , @xmath138 alternatively , the amplitude can be expressed in terms of the generalized ( complex ) sachs electric and magnetic form factors , @xmath139 and @xmath140 , in which case the reduced cross section , up to order @xmath141 corrections , can be written @xcite @xmath142 where the form factor @xmath143 has been expressed in terms of the ratio @xmath144 with @xmath145 .
we should emphasize that the generalized form factors are not observables , and therefore have no intrinsic physical meaning .
thus the magnitude and @xmath20 dependence of the generalized form factors will depend on the choice of parametrization of the generalized amplitude .
for example , the axial parametrization introduces an effective axial vector coupling beyond born level , and is written as @xcite @xmath146 following ref .
@xcite , one finds the relationships @xmath147 in fig .
[ fig : df ] we show the contributions of 2@xmath34 exchange to the ( real parts of the ) proton @xmath148 and @xmath149 form factors , and the ratio @xmath150 evaluated at @xmath110 , 3 and 6 gev@xmath26 .
one observes that the 2@xmath34 correction to @xmath149 is large , with a positive slope in @xmath20 which increases with @xmath3 .
the correction to @xmath148 is similar to that for @xmath149 at @xmath114 gev@xmath26 , but becomes shallower at intermediate @xmath20 values for larger @xmath3 .
both of these corrections are significantly larger than the @xmath150 correction , which is weakly @xmath3 dependent , and has a small negative slope in @xmath20 at larger @xmath3 .
the contribution to @xmath150 is found to be about 5 times smaller than that extracted in phenomenological analyses @xcite under the assumption that the entire form factor discrepancy is due to the new @xmath143 contribution ( see also ref .
@xcite ) . direct experimental evidence for the contribution of 2@xmath34 exchange can be obtained by comparing @xmath151 and @xmath152 cross sections through the ratio @xmath153 whereas the born amplitude @xmath51 changes sign under the interchange @xmath154 , the 2@xmath34 exchange amplitude @xmath72 does not .
the interference of the @xmath51 and @xmath72 amplitudes therefore has the opposite sign for electron and positron scattering .
since the finite part of the 2@xmath34 contribution is negative over most of the range of @xmath20 , one would expect to see an enhancement of the ratio of @xmath155 to @xmath156 cross sections , @xmath157 where @xmath109 is defined in eq .
( [ eq : delta_dif ] ) . although the current data on elastic @xmath7 and @xmath6 scattering are sparse , there are some experimental constraints from old data taken at slac @xcite , cornell @xcite , desy @xcite and orsay @xcite ( see also ref .
the data are predominantly at low @xmath3 and at forward scattering angles , corresponding to large @xmath20 ( @xmath158 ) , where the 2@xmath34 exchange contribution is small ( @xmath159 ) .
nevertheless , the overall trend in the data reveals a small enhancement in @xmath160 at the lower @xmath20 values , as illustrated in fig .
[ fig : ree ] ( which shows a subset of the data , from the slac experiments @xcite ) . the data in fig .
[ fig : ree ] are compared with our theoretical results , calculated for several fixed values of @xmath3 ( @xmath110 , 3 and 6 gev@xmath26 ) .
the results are in good agreement with the data , although the errors on the data points are quite large .
clearly better quality data at backward angles , where an enhancement of up to @xmath161 is predicted , would be needed for a more definitive test of the 2@xmath34 exchange mechanism .
an experiment @xcite using a beam of @xmath162 pairs produced from a secondary photon beam at jefferson lab will make simultaneous measurements of @xmath7 and @xmath6 elastic cross sections up to @xmath163 gev@xmath26 . a proposal to perform a precise ( @xmath164 ) comparison of @xmath7 and
@xmath6 scattering at @xmath165 gev@xmath26 and @xmath166 has also been made at the vepp-3 storage ring @xcite .
the results of the 2@xmath34 exchange calculation in the previous section give a clear indication of a sizable correction to the lt - separated data at moderate and large @xmath3 . the obvious question which arises is whether , and to what extent , the 2@xmath34 exchange affects the polarization transfer results , which show the dramatic fall - off of the @xmath37 ratio at large @xmath3 . in this section
we examine this problem in detail .
the polarization transfer experiment involves the scattering of longitudinally polarized electrons from an unpolarized proton target , with the detection of the polarization of the recoil proton , @xmath167 .
( the analogous process whereby a polarized electron scatters elastically from a polarized proton leaving an unpolarized final state gives rise to essentially the same information . ) in the born approximation the spin dependent amplitude is given by @xmath168 where @xmath169 and @xmath170 are the spin four - vectors of the initial electron and final proton , respectively , and the spinor @xmath171 is defined such that @xmath172 , and similarly for @xmath173 . the spin four - vector ( for either the electron or recoil proton ) can be written in terms of the 3-dimensional spin vector @xmath174 specifying the spin direction in the rest frame ( see _ e.g. _ ref .
@xcite ) , @xmath175 where @xmath176 and @xmath177 are the mass and energy of the electron or proton . clearly in the limit @xmath178 , the spin four - vector @xmath179 .
since @xmath174 is a unit vector , one has @xmath180 , and one can verify from eq .
( [ eq : s ] ) that @xmath181 and @xmath182 .
if the incident electron energy @xmath29 is much larger than the electron mass @xmath176 , the electron spin four - vector can be related to the electron helicity @xmath183 by @xmath184 the coordinate axes are chosen so that the recoil proton momentum @xmath185 defines the @xmath186 axis , in which case for longitudinally polarized protons one has @xmath187 . in the 1@xmath34 exchange approximation the elastic cross section for scattering a longitudinally polarized electron with a recoil proton polarized longitudinally is then given by @xmath188 for transverse recoil proton polarization we define the @xmath189 axis to be in the scattering plane , @xmath190 , where @xmath191 defines the direction perpendicular , or normal , to the scattering plane .
the elastic cross section for producing a transversely polarized proton in the final state , with @xmath192 , is given by @xmath193 taking the ratio of the transverse to longitudinal proton cross sections then gives the ratio of the electric to magnetic proton form factors , as in eq .
( [ eq : poltrans ] ) .
note that in the 1@xmath34 exchange approximation the normal polarization is identically zero . the amplitude for the 2@xmath34 exchange diagrams in fig .
[ fig : diag ] with the initial electron and final proton polarized can be written as @xmath194 where the numerators are the matrix elements @xmath195 and the denominators are given in eqs .
( [ eq : dbox ] ) and ( [ eq : dxbox ] ) . the traces in eqs .
( [ eq : nbox_pol ] ) and ( [ eq : nxbox_pol ] ) can be evaluated using the explicit expression for the spin - vectors @xmath196 and @xmath197 in eqs .
( [ eq : s ] ) and ( [ eq : s1 ] ) . in analogy with the unpolarized case ( see eq .
( [ eq : delta_dif ] ) ) , the spin - dependent corrections to the longitudinal ( @xmath198 ) and transverse ( @xmath199 ) cross sections are defined as the finite parts of the 2@xmath34 contributions relative to the ir expression from mo & tsai @xcite in eq .
( [ eq : deltairmt ] ) , which are independent of polarization , @xmath200 experimentally , one does not usually measure the longitudinal or transverse cross section _ per se _ , but rather the ratio of the transverse or longitudinal cross section to the unpolarized cross section , denoted @xmath1 or @xmath2 , respectively .
thus the 2@xmath34 exchange correction to the polarization transfer ratio can be incorporated as @xmath201 where @xmath109 is the correction to the unpolarized cross section considered in the previous section .
the 2@xmath34 exchange contribution relative to the born term is shown in fig .
[ fig : dellt ] .
the correction to the longitudinal polarization transfer ratio @xmath1 is small overall .
this is because the correction @xmath198 to the longitudinal cross section is roughly the same as the correction @xmath109 to the unpolarized cross section .
the corrections @xmath109 and @xmath198 must be exactly the same at @xmath202 ( @xmath203 ) , and our numerical results bear this out . by contrast , the correction to the transverse polarization transfer ratio @xmath2 is enhanced at backward angles , and grows with @xmath3 .
this is due to a combined effect of @xmath199 becoming more positive with increasing @xmath3 , and @xmath109 becoming more negative . in the standard radiative corrections using the results of mo & tsai @xcite , the corrections for transverse polarization are the same as those for longitudinal polarization , so that no additional corrections beyond hard bremsstrahlung need be applied @xcite . because the polarization transfer experiments @xcite typically have @xmath2040.8 , the shift in the polarization transfer ratio in eq .
( [ eq : poltrans ] ) due to the 2@xmath34 exchange corrections is not expected to be dramatic . if @xmath27 is the corrected ( `` true '' ) electric to magnetic form factor ratio , as in eq .
( [ eq : sigrcor ] ) , then the measured polarization transfer ratio is @xmath205 inverting eq .
( [ eq : shiftedr ] ) , the shift in the ratio @xmath27 is illustrated in fig .
[ fig : gempt ] by the filled circles ( offset slightly for clarity ) .
the unshifted results are indicated by the open circles , and the lt separated results are labeled by diamonds . the effect of the 2@xmath34 exchange on the form factor ratio is a very small , @xmath4 suppression of the ratio at the larger @xmath3 values , which is well within the experimental uncertainties .
note that the shift in @xmath27 in eq .
( [ eq : shiftedr ] ) does not include corrections due to hard photon bremsstrahlung ( which are part of the standard radiative corrections ) .
since these would make both the numerator and denominator in eq .
( [ eq : shiftedr ] ) even larger , the correction shown in fig .
[ fig : gempt ] would represent an upper limit on the shift in @xmath27 .
finally , the 2@xmath34 exchange process can give rise to a non - zero contribution to the elastic cross section for a recoil proton polarized normal to the scattering plane .
this contribution is purely imaginary , and does not exist in the 1@xmath34 exchange approximation .
it is illustrated in fig .
[ fig : deln ] , where the ratio @xmath206 of the 2@xmath34 exchange contribution relative to the _ unpolarized _ born contribution is shown as a function of @xmath20 for several values of @xmath3 .
( for consistency in notation we denote this correction @xmath206 rather than @xmath207 , even though there is no ir contribution to the normal polarization . )
the normal polarization contribution is very small numerically , @xmath208 , and has a very weak @xmath20 dependence .
in contrast to @xmath198 and @xmath199 , the normal polarization ratio is smallest at low @xmath20 , becoming larger with increasing @xmath20 .
although not directly relevant to the elastic form factor extraction , the observation of protons with normal polarization would provide direct evidence of 2@xmath34 exchange in elastic scattering .
figure [ fig : ay ] shows the normal polarization asymmetry @xmath209 as a function of the center of mass scattering angle , @xmath210 , for several values of @xmath3 .
the asymmetry is relatively small , of the order of 1% at small @xmath210 for @xmath211 gev@xmath26 , but grows with @xmath3 .
the imaginary part of the 2@xmath34 amplitude can also be accessed by measuring the electron beam asymmetry for electrons polarized normal to the scattering plane @xcite .
knowledge of the imaginary part of the 2@xmath34 exchange amplitude could be used to constrain models of compton scattering , although relating this to the real part ( as needed for form factor studies ) would require a dispersion relation analysis .
in this section we examine the effect of the 2@xmath34 exchange contribution on the form factors of the neutron . since the magnitude of the electric form factor of the neutron is relatively small compared with that of the proton , and as we saw in sec .
iii the effects on the proton are significant at large @xmath3 , it is important to investigate the extent to which @xmath212 may be contaminated by 2@xmath34 exchange . using the same formalism as in secs .
ii and iii , the calculated 2@xmath34 exchange correction for the neutron is shown in fig .
[ fig : del_n ] for @xmath110 , 3 and 6 gev@xmath26 . since there is no ir divergent contribution to @xmath58 for the neutron , the total 2@xmath34 correction @xmath213
is displayed in fig .
[ fig : del_n ] . in the numerical calculation , the input neutron form factors from ref .
@xcite are parameterized using the pole fit in eq .
( [ eq:3pole ] ) , with the parameters given in table [ tab : param ] . for comparison ,
the correction at @xmath115 gev@xmath26 is also computed using a 3-pole fit to the form factor parameterization from ref .
the difference between these is an indication of the model dependence of the calculation .
the most notable difference with respect to the proton results is the sign and slope of the 2@xmath34 exchange correction .
namely , the magnitude of the correction @xmath214 for the neutron is @xmath215 times smaller than for the proton .
the reason for the sign change is the negative anomalous magnetic moment of the neutron .
the @xmath20 dependence is approximately linear at moderate and high @xmath20 , but at low @xmath20 there exists a clear deviation from linearity , especially at large @xmath3 . translating the @xmath20 dependence to the form factor ratio , the resulting shift in @xmath216
is shown in fig .
[ fig : gemn_lt ] at several values of @xmath3 , assuming a linear 2@xmath34 correction over two different @xmath20 ranges ( @xmath127 and @xmath128 ) . the baseline ( uncorrected ) data are from the global fit in ref .
the shift due to 2@xmath34 exchange is small at @xmath110 gev@xmath26 , but increases significantly by @xmath217 gev@xmath26 , where it produces a 5060% rise in the uncorrected ratio .
these results suggest that , as for the proton , the lt separation method is subject to large corrections from 2@xmath34 exchange at large @xmath3 . while the 2@xmath34 corrections to the form factor ratio from lt separation are signficant , particularly at large @xmath3 , in practice the neutron @xmath212 form factor
is commonly extracted using the polarization transfer method . to compare the 2@xmath34 effects on the ratio @xmath218 extracted by polarization transfer , in fig .
[ fig : gemn_pt ] we plot the same `` data points '' as in fig . [
fig : gemn_lt ] , shifted by the @xmath219 corrections as in eq .
( [ eq : shiftedr ] ) at two values of @xmath20 ( @xmath220 and 0.8 ) .
the shift is considerably smaller than that from the lt method , but nevertheless represents an approximately 4% ( 3% ) suppression at @xmath220 ( 0.8 ) for @xmath221 gev@xmath26 , and @xmath222 ( 5% ) suppression for @xmath115 gev@xmath26 for the same @xmath20 . in the jefferson lab experiment @xcite to measure @xmath223 at @xmath224 gev@xmath26 the value of @xmath20 was around 0.9 , at which the 2@xmath34 correction was @xmath225 . in the recently approved extension of this measurement to @xmath226
gev@xmath26 @xcite , the 2@xmath34 correction for @xmath227 is expected to be around 3% . while small , these corrections will be important to take into account in order to achieve precision at the several percent level .
in this section we extend our formalism to the case of elastic scattering from @xmath8he nuclei .
of course , the contribution of @xmath8he intermediate states in 2@xmath34 exchange is likely to constitute only a part of the entire effect
contributions from break - up channels may also be important .
however , we can obtain an estimate on the size of the effect on the @xmath8he form factors , in comparison with the effect on the nucleon form factor ratio .
the expressions used to evaluate the 2@xmath34 contributions are similar to those for the nucleon , since @xmath8he is a spin-@xmath228 particle , although there are some important differences . for instance , the charge is now @xmath229 ( where @xmath230 is the atomic number of @xmath8he ) , the mass @xmath231 is @xmath232 times larger than the nucleon mass , and the anomalous magnetic moment is @xmath233 . in addition , the internal @xmath234 form factor is somewhat softer than the corresponding nucleon form factor ( since the charge radius of the @xmath8he nucleus is @xmath235 fm ) . using a dipole shape for the form factor
gives a cut - off mass of @xmath236 gev .
the 2@xmath34 exchange correction is shown in fig .
[ fig : del_he3 ] as a function of @xmath20 for several values of @xmath3 .
the @xmath20 dependence illustrates the interesting interplay between the dirac and pauli contributions to the cross section . at low @xmath3 ( @xmath237 gev@xmath26 ) ,
the @xmath9 contribution is dominant , and the effect has the same sign and similar magnitude as in the proton .
the result in fact reflects a partial cancellation of 2 opposing effects : the larger charge squared @xmath238 of the @xmath8he nucleus makes the effect larger ( by a factor @xmath239 ) , while the larger mass squared of the @xmath8he nucleus suppresses the effect by a factor @xmath240 .
in addition , the form factor used is much softer than that of the nucleon , so that the overall effect turns out to be similar in magnitude as for the proton . with increasing @xmath3
the pauli @xmath10 term becomes more important , so that for @xmath241 gev@xmath26 the overall sign of the contribution is positive .
interestingly , over most of the region between @xmath242 and 0.9 the slope in @xmath20 is approximately constant .
this allows us to extract the correction to the ratio of charge to magnetic form factors , @xmath243 , which we illustrate in fig .
[ fig : r_he3 ] .
the effect is a small , @xmath244 reduction in the ratio for @xmath245 gev@xmath26 , which turns into an enhancement at large @xmath3 .
however , the magnitude of the effect is small , and even for @xmath246gev@xmath26 the 2@xmath34 effect only gives @xmath247 increase in the form factor ratio .
proposed experiments at jefferson lab @xcite would measure the @xmath8he form factors to @xmath248 gev@xmath26 .
we have presented a comprehensive analysis of the effects of 2@xmath34 exchange in elastic electron
nucleon scattering , taking particular account of the effects of nucleon structure .
our main purpose has been to quantify the 2@xmath34 effect on the ratio of electric to magnetic form factors of the proton , which has generated controversy recently stemming from conflicting results of measurements at large @xmath3 .
consistent with the earlier preliminary investigation @xcite , we find that inclusion of 2@xmath34 exchange reduces the @xmath37 ratio extracted from lt - separated cross section data , and resolves a significant amount of the discrepancy with the polarization transfer results . at higher @xmath3
we find strong deviations from linearity , especially at small @xmath20 , which can be tested in future high - precision cross section measurements .
there is some residual model - dependence in the calculation of the 2@xmath34 amplitude arising from the choice of form factors at the internal @xmath249 vertices in the loop integration .
this dependence , while not overwhelming , will place limitations on the reliability of the lt separation technique in extracting high-@xmath3 form factors . on the other hand ,
the size of the 2@xmath34 contributions to elastic scattering could be determined from measurement of the ratio of @xmath152 to @xmath151 elastic cross sections , which are uniquely sensitive to 2@xmath34 exchange effects .
we have also generalized our analysis to the case where the initial electron and recoil proton are polarized , as in the polarization transfer experiments .
while the 2@xmath34 corrections can be as large as @xmath2395% at small @xmath20 for @xmath250 gev@xmath26 , because the polarization transfer measurements are performed typically at large @xmath20 we find the impact on the extracted @xmath37 ratio to be quite small , amounting to @xmath4 suppression at the highest @xmath3 value .
extending the formalism to the case of the neutron , we have calculated the 2@xmath34 exchange corrections to the neutron @xmath223 ratio . while numerically smaller than for the proton , the corrections are nonetheless important since the magnitude of @xmath212 itself is small compared with @xmath117 .
furthermore , because of the opposite sign of the neutron magnetic moment relative to the proton , the 2@xmath34 corrections to the lt - separated cross section give rise to a sizable enhancement of @xmath223 at large @xmath3 .
the analogous effects for the polarization transfer ratio are small , on the other hand , giving rise to a few percent suppression for @xmath25 gev@xmath26 .
finally , we have also obtained an estimate of the 2@xmath34 exchange contribution to the elastic form factors of @xmath8he from elastic intermediate states .
the results reveal an interesting interplay between an enhancement from the larger charge of the @xmath8he nucleus and a suppression due to the larger mass . together with softer form factor ( larger radius ) compared with that of the nucleon , the net effect is @xmath159 over the @xmath3 range accessible to current and upcoming experiments .
contributions from excited states , such as the @xmath109 and heavier baryons , may modify the quantitative analysis presented here .
naively , one could expect their effect to be suppressed because of the larger masses involved , at least for the real parts of the form factors .
an investigation of the inelastic excitation effects is presented in ref .
@xcite .
we would like to thank j. arrington for helpful discussions and communications .
this work was supported in part by nserc ( canada ) , doe grant de - fg02 - 93er-40762 , and doe contract de - ac05 - 84er-40150 under which the southeastern universities research association ( sura ) operates the thomas jefferson national accelerator facility ( jefferson lab ) . ) ) and the commonly used expression ( [ eq : deltairmt ] ) from mo & tsai @xcite for @xmath1106 gev@xmath26 .
the numbers labeling the curves denote the respective @xmath3 values in gev@xmath26 .
[ fig : delta],height=453 ] , 6 and 12 gev@xmath26 , using realistic ( solid ) @xcite and dipole ( dashed ) form factors ; ( b ) at @xmath217 gev@xmath26 using the form factor parameterizations from refs
. @xcite ( solid ) , @xcite ( dashed ) , and @xcite with @xmath117 constrained by the lt - separated ( dot - dashed ) and polarization transfer ( long - dashed ) data .
[ fig : ffcomp],title="fig:",height=453 ] , 6 and 12 gev@xmath26 , using realistic ( solid ) @xcite and dipole ( dashed ) form factors ; ( b ) at @xmath217 gev@xmath26 using the form factor parameterizations from refs .
@xcite ( solid ) , @xcite ( dashed ) , and @xcite with @xmath117 constrained by the lt - separated ( dot - dashed ) and polarization transfer ( long - dashed ) data .
[ fig : ffcomp],title="fig:",height=453 ] ( scaled by the dipole form factor @xmath254 ) versus @xmath20 for several values of @xmath3 : ( a ) slac data @xcite at @xmath119 ( open squares ) , 4 ( filled circles ) , 5 ( open circles ) and 6 gev@xmath26 ( filled squares ) ; ( b ) jlab data @xcite at @xmath255 ( filled squares ) , 3.2 ( open squares ) and 4.1 gev@xmath26 ( filled circles ) . the dotted curves are born cross sections evaluated using a form factor parameterization @xcite with @xmath117 fitted to the polarization transfer data @xcite , while the solid curves include 2@xmath34 contributions .
the curves in the bottom panel have been shifted by ( + 1.0% , + 2.1% , + 3.0% ) for @xmath256 gev@xmath26 . [
fig : sigr],title="fig:",height=453 ] ( scaled by the dipole form factor @xmath254 ) versus @xmath20 for several values of @xmath3 : ( a ) slac data @xcite at @xmath119 ( open squares ) , 4 ( filled circles ) , 5 ( open circles ) and 6 gev@xmath26 ( filled squares ) ; ( b ) jlab data @xcite at @xmath255 ( filled squares ) , 3.2 ( open squares ) and 4.1 gev@xmath26 ( filled circles ) . the dotted curves are born cross sections evaluated using a form factor parameterization @xcite with @xmath117 fitted to the polarization transfer data @xcite , while the solid curves include 2@xmath34 contributions .
the curves in the bottom panel have been shifted by ( + 1.0% , + 2.1% , + 3.0% ) for @xmath256 gev@xmath26 . [
fig : sigr],title="fig:",height=453 ] measured using lt separation ( open diamonds ) @xcite and polarization transfer ( pt ) ( open circles ) @xcite .
the lt points corrected for 2@xmath34 exchange are shown assuming a linear slope for @xmath127 ( filled squares ) and @xmath128 ( filled circles ) ( offset for clarity ) .
[ fig : gemp],height=453 ] contributions ( defined with respect to the mo - tsai ir result @xcite ) to the real parts of the @xmath90 ( dashed ) , @xmath89 ( dot - dashed ) and @xmath150 ( solid ) form factors of the proton at @xmath110 , 3 and 6 gev@xmath26 .
note the larger scale in the bottom figure .
[ fig : df],title="fig:",height=340 ] contributions ( defined with respect to the mo - tsai ir result @xcite ) to the real parts of the @xmath90 ( dashed ) , @xmath89 ( dot - dashed ) and @xmath150 ( solid ) form factors of the proton at @xmath110 , 3 and 6 gev@xmath26 .
note the larger scale in the bottom figure .
[ fig : df],title="fig:",height=340 ] contributions ( defined with respect to the mo - tsai ir result @xcite ) to the real parts of the @xmath90 ( dashed ) , @xmath89 ( dot - dashed ) and @xmath150 ( solid ) form factors of the proton at @xmath110 , 3 and 6 gev@xmath26 . note the larger scale in the bottom figure .
[ fig : df],title="fig:",height=340 ] to @xmath152 cross sections .
the data are from slac @xcite , with @xmath3 ranging from 0.01 to 5 gev@xmath26 .
the results of the 2@xmath34 exchange calculations are shown by the curves for @xmath110 ( dotted ) , 3 ( dashed ) and 6 gev@xmath26 ( solid ) .
[ fig : ree],height=453 ] ) ) of the born+2@xmath34 correction relative to the born term , for ( a ) longitudinal and ( b ) transverse recoil proton polarization , at @xmath110 ( dotted ) , 3 ( dashed ) and 6 gev@xmath26 ( solid ) .
note the different scales on the vertical axes .
[ fig : dellt],title="fig:",height=453 ] + ) ) of the born+2@xmath34 correction relative to the born term , for ( a ) longitudinal and ( b ) transverse recoil proton polarization , at @xmath110 ( dotted ) , 3 ( dashed ) and 6 gev@xmath26 ( solid ) .
note the different scales on the vertical axes .
[ fig : dellt],title="fig:",height=453 ] contribution to the unpolarized electron neutron elastic scattering cross section , at @xmath114 ( dotted ) , 3 ( dashed ) and 6 gev@xmath26 ( solid and dot - dashed ) .
the dot - dashed curve corresponds to the form factor parameterization of ref .
@xcite , while the others are from ref .
( as fitted by the parameters in table i ) .
[ fig : del_n],height=453 ] exchange on the ratio of neutron form factors @xmath216 using lt separation .
the uncorrected points ( open circles ) are from the form factor parameterization in ref .
@xcite , while the points corrected for 2@xmath34 exchange are obtained from linear fits to @xmath213 in fig .
[ fig : del_n ] for @xmath127 ( filled squares ) and @xmath128 ( filled circles ) ( offset for clarity ) .
[ fig : gemn_lt],height=453 ] exchange on the ratio of neutron form factors @xmath216 using polarization transfer .
the uncorrected points ( open circles ) are from the parameterization in ref .
@xcite , and the points corrected for 2@xmath34 exchange correspond to @xmath220 ( filled squares ) and @xmath257 ( filled circles ) ( offset for clarity ) .
[ fig : gemn_pt],height=453 ] contribution to the unpolarized electron@xmath8he elastic scattering cross section , with the @xmath8he elastic intermediate state , as a function of @xmath20 , for @xmath110 ( dotted ) , 2 ( dot - dashed ) , 3 ( dashed ) and 6 gev@xmath26 ( solid ) .
[ fig : del_he3],height=453 ] | a detailed study of two - photon exchange in unpolarized and polarized elastic electron
nucleon scattering is presented , taking particular account of nucleon finite size effects .
contributions from nucleon elastic intermediate states are found to have a strong angular dependence , which leads to a partial resolution of the discrepancy between the rosenbluth and polarization transfer measurements of the proton electric to magnetic form factor ratio , @xmath0 .
the two - photon exchange contribution to the longitudinal polarization transfer @xmath1 is small , whereas the contribution to the transverse polarization transfer @xmath2 is enhanced at backward angles by several percent , increasing with @xmath3 .
this gives rise to a small , @xmath4 suppression of @xmath0 obtained from the polarization transfer ratio @xmath5 at large @xmath3 .
we also compare the two - photon exchange effects with data on the ratio of @xmath6 to @xmath7 cross sections , which is predicted to be enhanced at backward angles .
finally , we evaluate the corrections to the form factors of the neutron , and estimate the elastic intermediate state contribution to the @xmath8he form factors . |
we begin by considering the hamiltonian describing the optomechanical interaction between two distinct optical modes ( indexed @xmath7 ) coupled to a shared mechanical mode with annihilation operator @xmath8 , h & = & _ k _ k + _
m + ( + ) _ k g_k .
+ [ eqn : hamiltonian ] each optical mode has a frequency @xmath16 , and is driven by a laser at frequency @xmath14 . the hamiltonian above is written in the interaction picture with @xmath116 . as shown below , strong driving at a mechanical frequency red detuned from each cavity mode , @xmath117 ,
causes an effective beam splitter interaction to take place between the mechanical mode and each optical cavity mode at an enhanced coupling rate @xmath118 , where @xmath13 is the square root of the photon occupation in optical mode @xmath11 . as long as this coupling is weak with respect to the optical linewidths @xmath18 ( @xmath23 ) , an adiabatic elimination of the optical cavities results in new effective mechanical loss rates @xmath119 into each of the @xmath11 optical degrees of freedom in the system .
this `` loss '' can provide an effective coupling between the optical cavity modes by allowing the exchange of excitations between the optical resonances through the mechanical motion of the system .
we calculate exactly what these conversion rates are by using a scattering matrix formulation to understand the behaviour of the system .
though completely general expressions can be derived , to best understand the processes involved we focus on the parameter regime relevant to this experiment , i.e. the weak - coupling , sideband - resolved case where @xmath120 , and follow a scattering matrix derivation of the induced optomechanical coupling between the optical waveguides coupled to each cavity mode .
the linearized heisenberg - langevin equations of motion are found using equation ( [ eqn : hamiltonian ] ) , the displacement @xmath121 + , and the inclusion of the optical input signal ( @xmath122 ) , optical vacuum noise ( @xmath123 ) , and mechanical thermal fluctuation fields ( @xmath124 ) : ( t ) & = & -(i_m + ) - i_k g_k ( + ) - ( t ) + _ k(t ) & = & -(i_k + ) - ig_k(+ ) + & & - ( t ) - ( t ) .
the mechanical loss rate @xmath125 , determines the coupling of the system to the thermal bath , and provides the most significant contribution in terms of noise processes relevant to the performance of the optomechanical wavelength converter .
the other loss rates , @xmath18 , @xmath27 , and @xmath126 are , respectively for optical cavity mode @xmath11 , the total optical loss rate , the optical loss rate associated with coupling into the waveguide @xmath11 , and parasitic optical loss rate into all other channels that are undetected representing a loss of information . due to the local evanescent side - coupling scheme studied here ( which bi - directionally couples to the cavity mode ) , @xmath127 , i.e. only a fraction @xmath128 of the photons leaving the cavity get detected and @xmath129 .
the equations of motion are linear and thus the system can be analyzed more simple in the frequency domain . solving for the spectrum of the mechanical mode @xmath130 in terms of the input noise operators , we find : & ( ) = + & + _ k + & + _ k , where the mechanical frequency @xmath131 is now modified by the optical spring , and the mechanical linewidth is given by @xmath132 .
the optomechanical damping terms @xmath133 come from coupling of the mechanical system to the optical mode @xmath11 , and are given by the relation _
, k & & = 2|g_k|^2 + & & = _ k_m.[eqn : damping ] the last expression is a simplification that is often made , and is equivalent to looking at spectral properties at detunings from the mechanical frequency much smaller than @xmath134 ( so the optical lineshape is nt taken into account ) . under this approximation ,
the optical spectrum is ( ) & = + & + _ j + & + _ j + & - ( ) - ( ) .
[ eqn : a_w ] from this expression , we see that there are several noise operators incident on optical mode @xmath11 .
the thermal fluctuations from the environment , @xmath135 , are converted into noise photons over the mechanical bandwidth @xmath136 .
there is also an induced coupling to the optical mode @xmath137 , and photons originally incident only on mode @xmath138 , i.e. @xmath139 are now also coupled into @xmath6 .
finally we note that vacuum noise creation operators are also present in this expression .
these give rise to quantum noise through spontaneous emission of phonons , though we show below that as @xmath131 is made very large with respect to @xmath18 , these terms diminish in importance . from outside the optomechanical system
, we only have access to photons sent into the system , @xmath140 , and those leaving the system , @xmath141 , on transmission . the relation between these operators
is best understood through scattering parameters , and can be derived using the equation for @xmath142 ( eq . [ eqn : a_w ] ) and the input - output boundary conditions @xmath143 . after some algebra , the output operator is expressed as ( ) & = & s_,k ( ) ( ) + & & + t_k ( ) ( ) + s_kj ( ) ( ) + & & + _
m n_,m ( ) ( ) + & & + _ m s_,m ( ) ( ) + & & + _ m s_,m ( ) ( ) [ eqn : full_scattering_relation ] the scattering coefficient @xmath144 is the conversion efficiency of mechanical thermal noise to photons , and is given by s_,k ( ) = i .
the coefficient @xmath145 in our system is transmission amplitude , given by t_k ( ) = ( 1 - 2_k ) + _ k. in principle , this coefficient is the electromagnetically induced transparency transmission coefficient , though here it is written about the mechanical frequency for detunings on the order of @xmath136 and the @xmath134 s are too large ( @xmath146 ) for the optical lineshape to be considered in the expression .
this is the expected result of the weak coupling approximation .
finally , the most important coefficient is the wavelength conversion coefficient @xmath147 which is given by s_kj ( ) = .
[ eqn : skj_app ] we calculate the spectral density of the output field on port @xmath11 ( @xmath148 ) assuming an input field spectral density on the opposing optical port @xmath138 ( @xmath149 ) , vacuum inputs on all other optical channels , and thermal noise from a phonon bath with thermal occupation @xmath150 .
the spectral densities here have units of @xmath151 and can be interpreted as photon flux per unit bandwidth .
at first we ignore the field creation operators in the scattering relation ( [ eqn : full_scattering_relation ] ) ( which give rise to quantum noise and can be neglected when the mechanical resonator occupation number is greater than one ) and arrive at the following expression : s_,k ( ) & = & _ -^ ^ + & = & _ -^ s^_,k(- ) s_,k(^ ) + & & + t^_k(- ) t_k(^ ) + & & + s^_kj(- ) s_kj(^ ) + & & + n^_,k(- ) n_,k(^ ) + & & + n^_,j(- ) n_,j(^ ) ^. this reduces to , s_,k ( ) & = & _ k n_b + & & + _ k _ j s_,j(),accounting for the noise autocorrelation functions . from here , we see that the maximum photon wavelength conversion efficiency is given by _
= _ 1 _ 2 .
the conversion process has a bandwidth ( bw ) equal to the mechanical linewidth _
i + _ , 1 + _ , 2 . from the same expression , we calculate a value for the optical signal to noise ratio ( osnr ) , to find ^_kj = _ j s_,j ( ) .
[ eqn : osnr_classical ] equation ( [ eqn : osnr_classical ] ) only takes into account the thermal noise in the system and therefore holds only when @xmath152 . at higher powers ,
the quantum noise processes which give rise to the back - action limit in cooling also give rise to excess noise for wavelength conversion .
to consider these terms , the creation operators in eq .
( [ eqn : full_scattering_relation ] ) must not be neglected and the full relation is then found to be s_,k ( ) & = & _ k _ i n_b + & & + _ k _ , k ( ) ^2 + & & + _ k _ , j ( ) ^2 + & & + _ k _ j s_,j ( ) . the quantum limited @xmath153 is then found to be _ kj & = . for the quantum back - action terms to become significant as compared to the thermal noise on the mechanical system , one needs @xmath154 .
this is a regime that is yet to be reached in experiments and as such the quantum back - action noise terms can be neglected for experiments to date .
perhaps more importantly , we ask `` what is the amount of noise added to a signal of a single photon ? ''
. we can express the signal to noise ratios as @xmath155 , with n _
+ assuming @xmath156 , @xmath157 , and @xmath158 , this expression becomes n _ 2 ^ -1_j ( + ( ) ^2 ) . to achieve the threshold where a single photon incoming on port @xmath138 is converted , and is more likely to be detected on port @xmath11 than an upconverted thermal excitation , we require @xmath159 .
this requirement is equivalent to stating that in the absence of signals on port @xmath138 and @xmath11 , a cooled phonon occupation of @xmath160 must be achieved .
we note that @xmath161 does not make an appearance in this equation , since reduced output coupling efficiency attenuates the thermal noise and signal equivalently . for the same reason ,
@xmath162 is present in the noise quanta expression .
the factor of two can be understood to come from the fact that for good conversion , we require the photon to enter , and exit the system before a phonon is up converted .
the former two processes happen at @xmath163 , while the latter is given by the thermalization rate @xmath164 .
we begin by considering the input signal to the first optical cavity mode .
this signal is generated by weakly modulating the @xmath33 control beam ( represented by amplitude @xmath165 ) at a frequency @xmath63 using an electro - optic modulator ( eom ) producing a_1 & & a_1 + a_1 e^-i _ 1 t + a_1 e^i _ 1 te^i , [ eqn : a1 ] where @xmath166 is the modulation index of the input signal and @xmath167 accounts for any phase difference between the sidebands ( @xmath168 if there is pure amplitude modulation ) .
the input sideband at frequency @xmath169 is nearly resonant with the first cavity mode at @xmath170 , whereas the lower frequency sideband at @xmath171 is detuned by approximately @xmath172 . as such , only the upper frequency sideband is resonant with the transparency window of the first cavity mode , and only this sideband is converted by the optomechanical crystal cavity into a sideband at the second , output cavity mode @xmath38 .
the converted sideband is generated from the control laser beam @xmath35 of the second cavity mode with frequency @xmath173 .
thus , the output field amplitude near the frequency of the second control laser beam ( amplitude @xmath174 ) is , ignoring @xmath167 , a_2 a_2 + s_21(_1 ) e^i _ 1 t. the power measured by a photodetector from the optical signal eminating from the output cavity mode is proportional to p_2 = |a_2|^2 ( 1 + |s_21(_1)| _ 1 t ) .
we thus define @xmath175 as the modulation index of the output signal , p_2 = |a_2|^2 ( 1 + _ 2 _ 1 t ) . by careful calibration of @xmath166 and @xmath176 , and measurement of the control beam power ( @xmath177 ) , the conversion scattering matrix element can be determined : @xmath178 the appearance of @xmath167 in equation ( [ eqn : a1 ] ) , along with its sensitivity to effects such as chromatic dispersion in the optical fiber and components , make calibration of @xmath166 more challenging due to interference between the upper and lower frequency sidebands .
direct photodetection of the optical input signal results in a @xmath167-dependent measured beat signal between the carrier and the two sidebands . to bypass this problem
, we generate an additional single sideband using an acousto - optic modulator ( aom ) , and beat it against each of the input optical sidebands .
this is accomplished by splitting off a portion of the optical input signal prior to creating the eom sidebands , and frequency shifting it by @xmath179 ( see fig . [
fig : si_setup ] ) .
this aom - shifted signal is then recombined with the main input signal , giving an overall signal equal to a_,1 = a_1 + a_1 e^-i _ 1 t + a_1 e^i ( _ 1 t + ) + a_3 e^i _
t , where @xmath180 is the field amplitude of the signal split off from the input signal prior to the eom .
the total photodetected signal is then given by |a_,1|^2&= & |a_1|^2 + |a_3|^2 + & & + |a_1|^2 _ 1 ( _ 1 t + /2 ) ( /2 ) + & & + 2 |a_1 a_3| ( _ t ) + & & + |a_1 a_3| + & & + |a_1 a_3| . from this equation
we see that the optical power at the detector consists of components at zero frequency ( dc ) as well as four modulated tones ( see fig .
[ fig : si_1450_calib ] ) . by taking the ratio of the tone at @xmath181 and @xmath182 ,
the value of @xmath166 can be accurately determined , independent of @xmath167 .
note , that in order to determine @xmath176 no such additional calibration signal is required .
the filtering properties of the input and output cavities results in a single output sideband , with no phase dependence of the photodetected intensity spectrum . in this case ,
careful optical and electronic calibration of the measured output cavity ( @xmath38 ) optical transmission intensity , similar to that described in the supplementary information of ref .
@xcite , provides an accurate estimate of @xmath176 and the wavelength converted signal strength . .
the dashed black line represents the theoretical values from the nominal system parameters while the solid red line takes into account the effect of the large intracavity photon population of cavity mode 2 ( @xmath87 ) .
the blue circles are the extracted data points and show much better correspondence to theory when this effect is accounted for . ] to model the wavelength conversion process , all of the optical cavity mode and mechanical resonator parameters entering into eq .
[ eqn : skj_app ] must be independently determined .
most of these parameters are measured as described in the main text ; however , additional measurements were performed to determine the power dependent optical cavity loss in the silicon optomechanical crystal device .
due to two - photon absorption in silicon , the parasitic optical cavity loss of both cavity modes is not static , but rather depends upon both control beam powers .
this effect is particularly acute for the higher @xmath107-factor first - order cavity mode ( @xmath37 ) , where in fig .
[ fig : si_t0]a we plot the normalized on - resonant optical transmission ( @xmath183 ) of the first - order cavity mode versus the power ( as represented by the intra - cavity photon population , @xmath87 ) of control beam @xmath35 feeding the second - order cavity mode .
the steady rise in the on - resonance transmission versus @xmath87 is attributable to the increased optical absorption loss stemming from free - carriers generated by two - photon absorption of the control beam @xmath35 . the corresponding change in coupling efficiency , @xmath184 ,
is shown on the right - side axis of fig .
[ fig : si_t0]a .
neglecting the effects of nonlinear optical absorption and free carriers on the intrinsic optical quality factor of @xmath37 leads to a theoretical model of conversion efficiency denoted by the dashed black line in fig [ fig : si_t0]b . taking into the account the nonlinear optical loss measured in fig .
[ fig : si_t0]a leads to the theoretical conversion efficiency denoted by the solid red line , showing much better correspondence with the data ( the solid red line is the model fit shown in the main text ) .
the effect of two - photon absorption and free - carrier absorption is much smaller on the second - order cavity mode ( @xmath38 ) due to its already lower optical @xmath107-factor , and we neglect it here .
the control - beam - generated free carriers also affect the intrinsic mechanical quality factor ( @xmath185 ) as was shown in ref .
@xcite , however , this affect is small for the control beam powers studied here ( the efficiency curve of fig [ fig : si_t0]b is also not modified as we plot it versus the ratio of cooperativities for which @xmath185 cancels out ) . | we theoretically propose and experimentally demonstrate coherent wavelength conversion of optical photons using photon - phonon translation in a cavity - optomechanical system .
for an engineered silicon optomechanical crystal nanocavity supporting a @xmath0 ghz localized phonon mode , optical signals in a @xmath1 mhz bandwidth are coherently converted over a @xmath2 thz frequency span between one cavity mode at wavelength @xmath3 nm and a second cavity mode at @xmath4 nm with a 93% internal ( 2% external ) peak efficiency . the thermal and quantum limiting noise involved in the conversion process
is also analyzed , and in terms of an equivalent photon number signal level are found to correspond to an internal noise level of only 6 and @xmath5 quanta , respectively .
the interaction of light with acoustic and molecular mechanical vibrations enables a great many optical functions used in communication systems today , such as amplification , modulation , wavelength conversion , and switching @xcite .
conventionally , these functions are realized in long ( centimeter to many meter ) waveguide based devices , relying on inherent materials properties and requiring intense optical pump beams . with the technological advancements in the fields of nanomechanics and nanophotonics , it is now possible to engineer interactions of light and mechanics .
progress in this area has included enhanced nonlinear optical interactions in structured silica fibers @xcite , near quantum - limited detection of nanomechanical motion @xcite , and the radiation pressure cooling of a mesoscopic mechanical resonator to its quantum ground state of motion @xcite .
coupling of electromagnetic and mechanical degrees of freedom , in which the coherent interaction rate is larger than the thermal decoherence rate of the system , as realized in the ground - state cooling experiments , opens up an array of new applications in classical and quantum optics .
this realization , along with the inherently broadband nature of radiation pressure , has spawned a variety of proposals for converting between photons of disparate frequencies @xcite through interaction with a mechanical degree of freedom - proposals which are but one of many possible expressions of hybrid optomechanical systems @xcite .
the ability to coherently convert photons between disparate wavelengths has broad technological implications not only for classical communication systems , but also future quantum networks @xcite .
for example , hybrid quantum networks require a low loss interface capable of maintaining quantum coherence while connecting spatially separate systems operating at incompatible frequencies @xcite . for this reason , photons operating in the low loss telecommunications band
are often proposed as a conduit for connecting different physical quantum systems @xcite .
it has also been realized that a wide variety of quantum systems lend themselves to coupling with mechanical elements . a coherent interface between mechanics and optics , then ,
could provide the required quantum links of a hybrid quantum network @xcite . until now
, most experiments demonstrating either classical or quantum wavelength conversion have used intrinsic optical nonlinearities of materials @xcite . in this letter
we show that by patterning an optomechanical crystal ( omc ) nanobeam from a thin film of silicon , engineered ghz mechanical resonances can be used to convert photons within a @xmath1 mhz bandwidth coherently over an optical frequency span of 11.2 thz , at internal efficiencies exceeding 90% , and with a thermal(quantum)-limited noise of only 6 ( @xmath5 ) quanta .
such cavity optomechanical systems are not just limited to the optical frequency domain , but may also find application to the interconversion of microwave and optical photons@xcite , enabling a quantum - optical interface to superconducting quantum circuits@xcite . conceptually , the proposed system for wavelength conversion can be understood from fig .
[ fig : concept]a . two optical cavity modes @xmath6 ( @xmath7 )
are coupled to a common mechanical mode @xmath8 via an interaction hamiltonian @xmath9 , where @xmath6 , @xmath8 are the annihilation operators and @xmath10 is the optomechanical coupling rate between the mechanical mode and the @xmath11th cavity mode .
physically , @xmath12 represents the frequency shift of cavity mode @xmath11 due to the zero - point motion of the mechanical resonator .
wavelength conversion is driven by two control laser beams ( @xmath13 in fig .
[ fig : concept]b ) , of frequency @xmath14 and nominal detuning @xmath15 to the red of cavity resonance at frequency @xmath16 . in the resolved sideband regime , where @xmath17 ( @xmath18 the bandwidth of the @xmath11th cavity mode ) , the spectral filtering of each cavity preferentially enhances photon - phonon exchange .
the resulting beam - splitter - like hamiltonian is @xmath19 @xcite , where @xmath20 is the parametrically enhanced optomechanical coupling rate due to the @xmath21 control beam ( @xmath22 is the control - beam induced intracavity photon number ) . in the weak - coupling limit , @xmath23 , this interaction effectively leads to an additional mechanical damping rate , @xmath24 .
the degree to which this optomechanical loss rate dominates the intrinsic mechanical loss is called the cooperativity , @xmath25 . at large cooperativities ,
the optomechanical damping has been used as a nearly noiseless loss channel to cool the mechanical mode to its ground state @xcite . in the case of a single optical cavity system
, it has also been used as a coherent channel allowing inter - conversion of photons and phonons leading to the observation of electromagnetically - induced transparency ( eit ) @xcite . in the double optical cavity system
presented here , coherent wavelength conversion of photons results . as shown in fig .
[ fig : concept]a , each optical cavity is coupled not just to a common mechanical mode , but also to an optical bath at rate @xmath26 and to an external photonic waveguide at rate @xmath27 ( the total cavity linewidth is @xmath28 ) .
the external waveguide coupling provides an optical interface to the wavelength converter , and in this work consists of a single - transverse - mode waveguide , bi - directionally coupled to each cavity mode .
the efficiency of the input / output coupling is defined as @xmath29 , half that of the total bi - directional rate .
although the wavelength converter operates symmetrically , here we will designate the higher frequency cavity mode ( @xmath30 ) as the input cavity and the lower frequency cavity ( @xmath31 ) as the output cavity .
as shown in fig .
[ fig : concept]b , photons sent into the wavelength converter with detuning @xmath32 from the control laser @xmath33 , are converted to photons @xmath34 detuned from the control laser @xmath35 , an @xmath2 thz frequency span for the device studied here .
the details of the conversion process can be understood by solving the heisenberg - langevin equations ( see appendix ) .
we linearize the system and work in the frequency domain , obtaining through some algebra the scattering matrix element @xmath36 , which is the complex , frequency - dependent conversion coefficient between the input field at cavity @xmath37 , and the output field at cavity @xmath38 .
this coefficient is given by the expression [ eqn : skj ] s_21 ( ) = , where @xmath39 is the total mechanical damping rate and equal to the bandwidth of the conversion process . from this expression , the spectral density of a converted signal @xmath40 ,
given the input signal spectral density @xmath41 , may be found and is given by s_,2 ( ) & = & _ 2 _ 1 ( n _ + s_,1 ( ) ) .
[ eqn : psd_expr ] these spectral densities have units of photons / hz@xmath42s and are proportional to optical power .
the added noise , @xmath43 , arises from thermal fluctuations of the mechanical system and the quantum back - action noise of light present in each optical mode . from here
, we see that in a system with ideal cavity - waveguide coupling , ( @xmath44 ) , the peak internal photon conversion efficiency is given by _ = .
this efficiency only depends on the internal coupling of the optomechanical system , and for both @xmath45 and @xmath46 , approaches unity .
the latter condition can be understood from requiring the coupling between the optical and mechanical modes to overtake the intrinsic mechanical loss rate , while the first requirement is due to impedance - like matching @xcite .
the total system efficiency is @xmath47 . the optomechanical system used in this work , shown in fig . [ fig : concept]c - f , consists not of a separate set of optical cavities , but rather of a single optomechanical crystal ( omc ) nanobeam cavity .
the omc nanobeam is fabricated from a silicon - on - insulator ( soi ) microchip @xcite , in which the top , @xmath48 nm thick , silicon device layer is patterned with a quasi - periodic linear array of etched air holes .
the larger air holes on either end of the beam induce bragg - like reflection of both guided optical and acoustic waves , resulting in strongly confined optical and mechanical resonances at the beam s center .
the device used in this work is designed to have both a first - order ( @xmath49 ) and a second - order ( @xmath50 ) optical cavity resonance of high quality factor .
both of these optical cavity modes are dispersively coupled to the same ghz - frequency mechanical resonance , depicted in fig .
[ fig : concept]f ( @xmath51 at @xmath52 ) .
a tapered optical fiber waveguide , placed in close proximity ( @xmath53 nm ) to the omc nanobeam cavity @xcite using precision stages , is used to evanescently couple light into and out of the cavity modes .
a schematic of the experimental set - up used to characterize the omc wavelength converter is shown in fig .
[ fig : eit]a .
measurements were performed in vacuum and at low temperature inside a continuous flow cryostat with a cold finger temperature of @xmath54 k ( the corresponding sample temperature , as inferred from the thermal bath temperature of the localized mechanical mode of the omc cavity , is @xmath55 k @xcite ) .
initial characterization of the optical cavity modes is performed by scanning the tunable control lasers across a wide bandwidth and recording the transmitted optical intensity through the optical fiber taper waveguide .
such a wavelength scan is shown in fig .
[ fig : eit]b , from which the resonance frequency ( @xmath56 , @xmath57 ) , cavity linewidth ( @xmath58 mhz , @xmath59 ghz ) , and waveguide coupling efficiency ( @xmath60 , @xmath61 ) of the two nanobeam cavity modes are determined
. further characterization of the optomechanical cavity is performed by using the control laser beams in conjunction with a weak sideband probe .
with control beams @xmath33 and @xmath35 detuned a mechanical frequency to the red of their respective cavity modes ( @xmath62 ) , a weak sideband signal generated from @xmath33 is swept across the first - order cavity mode .
the resulting reflected sideband signal versus sideband frequency shift @xmath63 is plotted in fig .
[ fig : eit]c , showing the broad cavity resonance along with a narrow central reflection dip ( see fig . [
fig : eit]c inset ) .
the narrow reflection dip , akin to the eit transparency window in atomic systems , is due to the interference between light coupled directly into the cavity mode and light coupled indirectly through the mechanical mode @xcite .
a similar eit response is shown in fig . [ fig : eit]d for the second - order cavity mode .
each of the transparency windows occur at @xmath64 , with a bandwidth equal to the optically damped mechanical linewidth ( @xmath65 ) . by fitting the optical resonance lineshape and transparency windows to theory
, one can also extract the control beam detunings @xmath66 . in what follows we use this sort of eit reflection spectroscopy , time - multiplexed in between wavelength conversion measurements , to set and stabilize the frequency of the control beams to @xmath67 . .
the calibration curve for cavity mode @xmath38 is performed with @xmath33 turned off , whereas the curve for cavity mode @xmath37 is generated with a weak @xmath35 such that @xmath68 for all measured points . *
b * , end - to - end power conversion efficiency of input signal to output signal for @xmath69 as a function of the ratio of the cooperativities of the control beams . in this plot
the control beam intensity for the first - order cavity mode is held fixed ( with @xmath70 ) , while the intensity of the control beam of the second - order cavity mode is swept from @xmath71 to @xmath72 .
the blue circles correspond to measured data points , whereas the solid red line is a theoretical curve using independently measured system parameters .
inset shows the conversion efficiency versus input signal detuning for matched control beams @xmath73 , indicating a conversion bandwidth of @xmath74 .
* c * , measured power spectrum of the output optical channel for @xmath75 , showing a series ( @xmath63 sweep ) of converted input tones ( green ) sitting on top of a much smaller thermal noise pedestal ( blue ) .
inset shows a zoom - in of the thermal noise pedestal . ]
coherent wavelength conversion can be thought of as occurring between the input transparency window of @xmath76 and the output transparency window of @xmath77 , with the phonon transition mediating the conversion .
as shown above , conversion efficiency is theoretically optimized for matched cooperativities of control beam @xmath33 and @xmath35 ( or equivalently @xmath78 ) .
calibration of the optically induced mechanical damping by the control beams can be performed as in the eit spectroscopy described above , or by measuring the spectral content of the photodetected transmission intensity of @xmath77 in the absence of any optical input .
such a measurement ( see set - up in fig .
[ fig : eit]a and appendix ) measures the mechanical resonator linewidth through the noise spectrum generated by the beating of noise sideband photons generated by thermal motion of the mechanical resonator with the control beam @xmath35 .
figure [ fig : results]a plots the inferred optomechanically induced damping of the mechanical resonator as a function of the power ( intracavity photon number , @xmath79 ) of control beams @xmath33 and @xmath35 , the slope of which gives the zero - point optomechanical coupling for both cavity modes ( @xmath80 and @xmath81 ) . to quantify the efficiency of the wavelength conversion process
the input cavity control beam @xmath33 is held fixed at a detuning @xmath82 and a power producing an intracavity photon population of @xmath83 , corresponding to large cooperativity @xmath70 .
the input signal , @xmath84 , is generated as an upper sideband of @xmath33 using an electro - optic intensity modulator ( the lower sideband is rejected by the wavelength converter as it is detuned to @xmath85 ) .
conversion of the @xmath84 sideband to an output signal emanating from cavity mode @xmath38 is completed by applying control beam @xmath35 with detuning @xmath86 .
the converted output signal sideband emitted from cavity @xmath38 is beat against the transmitted control beam @xmath35 on a high - speed photodiode .
the amplitude of the input signal tone near resonance with cavity mode @xmath37 is calibrated using a second reference signal generated through acousto - optic modulation of @xmath33 , whereas the amplitude of the output signal sideband is inferred from calibration of the control beam intracavity number @xmath87 and the optical transmission and detection chain ( see fig .
[ fig : eit]a and appendix ) . figure [ fig : results]b plots the resulting power conversion efficiency , given by the magnitude squared of eq .
( [ eqn : psd_expr ] ) , versus the ratio of @xmath88 as the power of control beam @xmath35 is varied and for an input signal exactly resonant with cavity mode @xmath37 ( @xmath89 ) .
the power conversion is referred directly to the input and output of the optomechanical cavity ( i.e. , not including additional losses in the optical link ) .
the solid red line shows the theoretical model for the conversion efficiency using the independently measured system parameters , and taking into account power - dependent effects on the optical losses in the silicon cavity modes ( see appendix ) .
good correspondence is seen between the measured data and theory , with a maximum of the conversion occurring at @xmath90 as expected from the matching condition .
the bandwidth of the wavelength conversion process is also probed , using matched conditions ( @xmath91 ) , and as above , with control beam detunings @xmath67 .
the modulation frequency generating @xmath84 is now varied from @xmath92 to @xmath93 , sweeping the narrow - band input sideband signal across the transparency window of cavity mode @xmath37 .
the corresponding measured output signal frequency shift follows that of the input signal ( @xmath94 ) , with the resulting power conversion efficiency plotted in the inset of fig .
[ fig : results]b .
a peak efficiency , measured from the input to the output port at the optomechanical cavity system , is @xmath95 , corresponding to an internal conversion efficiency of @xmath96 .
fitting the results to a lorentzian as in eq .
( [ eqn : skj ] ) , we find the conversion bandwidth to be @xmath97 , equal to the optically damped mechanical linewidth of the mechanical resonance , @xmath98 . due to the thermal bath and @xmath99 due to spontaneous stokes scattering from the two control beams .
cooling of the mechanical mode is also performed by the control beams , resulting in a theoretically cooled phonon occupancy of @xmath100 .
the output noise power is proportional to the output cavity cooling rate , @xmath101 .
* b * , output noise power spectral density . the measured noise data ( red circles )
is shown along with several theoretical noise curves .
the dashed red curve is the corresponding output noise power in the absence of control beam cooling .
the solid ( dashed ) green curve is the output noise power due to spontaneous stokes scattering in the presence ( absence ) of cavity filtering . ]
the spectrum of the converted output signal for a series of signal frequencies @xmath63 are shown in fig .
[ fig : results]c . as can be seen from this plot and the zoomed - in inset
, the narrow - band converted photons sit atop a noise pedestal of bandwidth corresponding to the damped mechanical resonator .
this noise is quantified by the added noise quanta number component of eq .
( [ eqn : psd_expr ] ) .
as shown in the appendix , under matched conditions the theoretical added noise is , n _ = 2 ^ -1_1 ( + ( ) ^2+()^2),[eqn : nadded ] where the first term arises due to thermal noise of the cooled and damped mechanical resonator , and the last two terms are quantum noise resulting from the spontaneous scattering of the control beams ( quantum back - action noise @xcite ) .
figure [ fig : noise]a pictorially indicates the various cooling , heating , and spontaneous scattering mechanisms that lead to the output noise of @xmath38 . a plot showing the measured output noise spectral density , calibrated in units of photon number ,
is given in figure [ fig : noise]b for the optimal conversion efficiency of fig .
[ fig : results]b . from the peak of the output noise spectral density ( @xmath102 ) in this plot ,
the added noise referred to the input of @xmath37 is estimated to be @xmath103 quanta .
the corresponding internal added noise ( for @xmath104 ) is only @xmath105 , predominantly due to the cooled phonon occupation of the mechanical resonator ( @xmath106 ) .
due to the strong filtering provided by the high-@xmath107 cavities ( @xmath108 ) , the quantum back - action noise contributes an insignificant @xmath109 ( @xmath110 ) added input ( internal ) noise photons . the effectiveness of noise reduction in the optomechanical cavity system is highlighted in fig .
[ fig : noise]b by showing the estimated thermal output noise in the absence of sideband cooling by the control beams ( dashed red curve ) and the quantum back - action noise in absence of cavity filtering ( green dashed curve ) .
although the demonstrated wavelength conversion in this work represents an important proof of principle , there are still considerable practical challenges to utilizing such cavity optomechanical devices for classical or quantum wavelength translation @xcite . with a thermal phonon occupancy @xmath111 required for single photon conversion at unity signal - to - noise , a primary concern for quantum networking @xcite
is the thermal energy stored in the mechanical resonator .
although resolved sideband cooling of nanomechanical resonators to occupancies below one has recently been demonstrated @xcite , this relation emphasizes the additional need for efficient cavity coupling . for chip - scale photonic devices similar to the one presented here
, the technology already exists for efficient optical connectivity , with optical fiber - to - chip coupling of @xmath112 @xcite and waveguide - to - cavity coupling efficiency of @xmath113 having been realized @xcite .
further reduction in the output noise may also be realized by simply reducing the temperature of operation ; for the @xmath0 ghz phonon frequency of the device in this study , @xmath114 for a bath temperature of @xmath115 mk realizable in a helium dilution refrigerator .
integration of cavity optomechanical devices into millikelvin experiments is particularly relevant for superconducting quantum circuits , which themselves have already been strongly couped to mechanical resonators @xcite .
addition of an optical cavity coupled directly , or via an additional acoustic waveguide , to the same mechanical resonator would provide a microwave - to - optical quantum interface @xcite similar to the optical - to - optical interface shown here . in the case of solid - state qubits with optical transitions
far from the telecommunication bands , such as the nitrogen - vacancy ( nv ) electron spin in diamond @xcite , the radiation pressure nonlinearity of cavity optomechanics could be substituted for intrinsic materials nonlinearities in proposed cavity - based schemes for single photon wavelength conversion and pulse shaping @xcite . due to the excellent mechanical properties of diamond and recent advancements in thin - film diamond photonics @xcite
, a cavity - optomechanical wavelength converter could be fabricated around an nv qubit from the same diamond host .
_ note : _ during the final preparation of this manuscript a similar worked appeared on the arxiv @xcite .
this work was supported by the darpa / mto meso program , institute for quantum information and matter , an nsf physics frontiers center with support of the gordon and betty moore foundation , and the kavli nanoscience institute at caltech .
jc and asn gratefully acknowledge support from nserc .
# 1 30 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) .
, , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , ,
* * , ( ) . ,
* * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , p. ( ) .
, , , , , * * , ( ) . , * * , ( ) , http://www.nature.com/nature/journal/v453/n7198/full/nature07127.html . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.106.130501 . , , , , , , , , , , * * , ( ) .
, , , , , * * , ( ) . , * * , ( ) . , , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , ,
* * , ( ) . , , , , * * , ( ) .
, , , , , , , * * , ( ) . , , , * * , ( ) .
, , , , , , , , , , , * * , ( ) .
, , , , , , , , * * , ( ) . , , , , , * * , ( ) . ,
, , , , * * , ( ) . , , , , , ( ) . |
cosmological @xmath4 body simulations are the primary instrument used to make theoretical predictions for structure formation in current models of the universe
. analytical understanding of the non - linear regime , crucial for many non - trivial tests of these models , remains very poor despite the many phenomenological results derived in simulations .
better understanding of the physics of non - linear clustering could help to better constrain and control cosmological simulations , which despite their ever increasing power and sophistication are still subject to questions about their accuracy and resolution . in this work we use
simplified one dimensional ( 1d ) toy models " of the full three dimensional ( 3d ) problem to study non - linear structure formation .
more specifically we study , using a family of such models , how the properties of non - linear clustering depend on both initial conditions and cosmology .
our main result is that , for cold initial conditions , we observe that the large phase space of initial conditions and cosmology we explore breaks up into two parts : on the one hand , a region in which there is an apparent `` universality '' of the non - linear clustering , i.e. , in which it depends very weakly both on initial conditions and on the cosmological expansion ; and on the other hand , a region where the non - linear clustering depends strongly on both the initial conditions and expansion history . in the latter region ,
the exponent characterizing the two point correlations is in excellent agreement with a simple analytical prediction which follows from the so - called stable clustering hypothesis , appropriately generalized to this class of models .
further the apparently well localized boundary dividing the phase space into these two disjoint regions can in fact be characterized using the stable clustering prediction , in terms of a critical value of the corresponding exponent .
the non - linear clustering appears to be strictly universal in the limit in which there is no expansion , but also very well approximated for steeper initial spectra when there is expansion .
we compare our results with those in three dimensions , and discuss how our results may provide a framework for understanding the much discussed apparent universality " of halo profiles in this context . the exploration of 1d models as a tool for understanding the physics of non - linear structure formation in a cosmological context goes back at least as far as the work of @xcite for the case of hot dark matter models .
different initial conditions and variants of the model have been discussed by a number of other authors @xcite .
there is also an extensive literature on the statistical mechanics of finite one dimensional self - gravitating systems ( see e.g. @xcite and references therein ) , as well as some studies which use this case to explore issues in cosmological n body simulations : @xcite uses it to probe discreteness effects , and @xcite the issue of universality in cold collapse .
the study reported here is a direct continuation of that in @xcite .
this latter paper showed , for the three 1d models studied in the previous literature , the very strong qualitative similarities in the temporal development of clustering to that in the 3d case .
for cold cosmological like initial conditions clustering is hierachical and driven by the linear amplification of fluctuations , leading , for the case of power law initial conditions , to self similar evolution of the non - linear correlations .
the exponent characterising the non - linear clustering was , further , found to be in very good agreement with that predicted by the so - called stable clustering " hypothesis . in this paper
we extend this study to a one parameter family of models which allows continuous interpolation between the static model and the two expanding models previously considered .
this allows us to probe fully the extent of validity of stable clustering , where it breaks down and what happens in this case .
the study reveals that there is a fairly abrupt switchover from the validity of stable clustering to an apparent universality in the clustering . in the next section
we define the class of 1d models we study , discussing also their relation to 3d cosmological models .
this presentation improves and simplifies that given in @xcite . for the family of scale - free models we focus on
, we derive , more rigorously than in @xcite , the prediction for the behaviour of an isolated structure and use it to derive the exponents predicted for the non - linear two point correlation when the stable clustering hypothesis is made . in the following section we discuss our numerical simulations , and in particular how we calibrate them using exact " simulations .
the next section gives our results , and in the last section we discuss their interpretation and possible relation to the three dimensional problem .
the class of 1d models we study are defined as follows .
we take the equations of motion _ in comoving coordinates _ of 3d cosmological @xmath4 body simulations , and simply replace the 3d newtonian gravitational interaction with the 1d gravitational interaction .
this gives 1d equations of motion @xmath5 where @xmath6 are the particle positions on the line .
the superscript ` j ' in the force term indicates that the sum , which extends over an infinite distribution of masses with non - zero mean mass density , is regularized , just as in 3d , by subtraction of the contribution of the mean mass density , i.e. , the force term is sourced only by fluctuations about the mean density . as we will discuss further below this sum
can be written explicitly in a simple manner .
the coupling @xmath7 is the 1d analogy of newton s constant ( multiplied by the particle mass ) . to establish a more direct relation between the 1d and 3d models
, one can consider the particles " in the 1d system to represent infinite parallel sheets of zero thickness embedded in 3d . in this case one
has the identification @xmath8 where @xmath9 is the surface mass density of the sheets .
the mean ( comoving ) mass density @xmath10 of the corresponding 3d universe and the number density @xmath11 of the 1d system are then related as @xmath12 , and therefore @xmath13 for any given 3d cosmological model which provides a scale factor @xmath14 and associated hubble constant @xmath15 we then have a _ unique _ corresponding 1d model .
the case @xmath16 ( and @xmath17 ) on the other hand , defines a model analogous to a 3d universe without expansion ( which is well defined but does not correspond to a cosmological model derived from general relativity ) .
the one parameter family of models we will study is a _ continuous interpolation between the 1d model obtained using the 3d eds cosmology , and the static model_. to define them precisely , and motivate their choice , it is convenient to change time variable in eq .
( [ 1deom1 ] ) defining @xmath18 .
this gives the equations of motion in the form @xmath19 i.e. , in which all effects of the cosmology appear only as a fluid type damping term , with @xmath20 for a 3d eds universe with comoving mass density @xmath10 , i.e. , @xmath21 where @xmath22 , we obtain a @xmath23 which is _ independent of time _ and given by @xmath24 where we have used eq .
( [ g - grelation ] ) to derive the last equality . in a @xmath25cdm cosmology ,
on the other hand , @xmath23 is approximately constant through the matter dominated era and increases slightly at late times .
we consider here simply the set of 1d models in which @xmath23 takes any constant value , and the focus of our study is the dependence of non - linear clustering on this parameter , as well as on the initial conditions .
@xmath23 can simply be considered as a control parameter for understanding the role of expansion in the determination of the properties of the non - linear clustering .
while most previous studies have considered either the eds model ( also known as the quintic " model for reasons which will be recalled below ) , or the static model , one other model in this family , corresponding to the case @xmath26 , introduced in @xcite , has been studied in @xcite as well as in @xcite .
we note , using eq .
( [ gamma - a - relation ] ) , that this class of models is obtained by taking the functional dependence of the 3d eds expansion law , but allowing a freedom in the normalisation of the expansion rate to the matter density , i.e. , we can obtain these models taking @xmath27 where @xmath2 is a positive constant .
( [ gamma - a - relation ] ) then still holds , and thus @xmath28 , but instead of ( [ gamma - eds ] ) we have , using eq . ( [ g - grelation ] ) , that @xmath29 in terms of 3d cosmological models the case @xmath30 , i.e. , @xmath31 , can be considered to be a model in which there is not just ordinary " matter but an additional pressureless component of the energy density which does not cluster , e.g. , a homogeneous scalar field with appropriate equation of state ( see e.g. @xcite and references therein ) .
equivalently one can consider that the @xmath32 models are obtained using the correct " normalisation for the hubble constant , @xmath33 , but choosing the 1d coupling freely instead of imposing the relation eq .
( [ g - grelation ] ) .
it is important to note that , given that 3d hubble law is imposed by hand " on the 1d model , there is in principle no correct value for @xmath23 .
our point of view here is simply to use @xmath23 ( or @xmath2 ) as a control parameter in trying to understand the role of expansion in non - linear structure formation . in other words
we are seeking to improve our _ qualitative _ understanding only through the study of these toy models .
in the context of the problem of structure formation the family of models we study has the interest of being , like eds models and static models in 3d , scale - free : the expansion introduces no additional time or length scales .
this has as a consequence that , for cold particles with initial density fluctuations with a power law spectrum , one expects to obtain , at sufficiently long times , temporal evolution which is self - similar " , i.e. , the evolution in time of the spatial correlation functions is equivalent to a simple time dependent rescaling of the spatial variables .
the reason why this occurs is that there is , in this case , only a single physically relevant length scale in the problem : the scale at which fluctuations are of order one , of which the evolution is predicted by linear theory .
further if the clustering in the strongly non - linear regime does not depend on this scale one would expect to obtain a scale - free behaviour of spatial correlations in this regime .
such power law behaviour has been observed in both expanding ( see , e.g. , @xcite and references therein ) and static 3d simulations @xcite for a range of power law initial conditions .
3d simulations , however , can show such behaviour over a very limited spatial range , making it difficult to establish if it is associated with a true scale invariance of the non - linear clustering .
the study of 1d models in the family we are considering @xcite , with much greater spatial resolution and numerical accuracy , shows very convincing evidence that these models do indeed give rise to power - law clustering indicative of truly scale - invariant clustering .
a central problem of non - linear structure formation is then that of understanding how the exponent ( or possibly exponents ) characterizing the non - linear clustering is determined . in 3d eds cosmology peebles ( see @xcite ) proposed many years ago a simple explanation of how such scale invariant clustering might arise : if highly non - linear structures ( of all different sizes ) are supposed to evolve essentially independently after their formation , they will virialize and remain stable in physical coordinates . from this
one may derive an exponent which depends only on the exponent in the power spectrum of initial density fluctuations .
we derive here now , more rigorously than the heuristic derivation given in @xcite , the prediction of such a stable clustering hypothesis for the 1d models we study .
to do so we simply derive and then analyse the equations of motion of a finite sub - system . in the simulations which we consider here , we start , as in 3d cosmological simulations , from a particle configuration generated by applying small displacements to an _
infinite _ perfect lattice configuration . at any moment in time
such a configuration can be fully specified by giving the displacements @xmath34 of particles from the sites of a perfect lattice .
the position of the @xmath35-th particle with respect to some arbitrary origin , is @xmath36 , where @xmath37 . in a particle labelling with @xmath38 for all @xmath35 ,
the displacements are uniquely defined , and such that no two particles cross when they are applied to the lattice . in this case
it has been shown @xcite that the force on the particle @xmath35 , given by the sum on the right side of eq .
( [ 1deom2 ] ) , with an appropriate specification of the regularisation , has the simple _ exact _ expression @xmath39 where @xmath40 is the average particle displacement .
this result has been shown for an infinite system with statistically translationally invariant displacements in @xcite , with the sole assumption of decay of correlations of the displacements at asymptotically large scales .
the same result has been shown to hold in @xcite for the case of an infinite _ periodic _ system , using an appropriate regularisation of an ewald summation . in this case , as we will now show explicitly , the expression ( [ 1d - exact - force - disp ] ) can be written as @xmath41 + 2gn_0[x_i - \langle x \rangle_p ] \label{1d - force - periodic}\ ] ] where @xmath42 ( @xmath43 ) is the number of particles in the periodic cell to the left ( right ) of particle @xmath35 , and @xmath44 is the average particle position , i.e. , centre of mass of the particles in the cell .
let us consider now any _ finite _ subsystem @xmath45 of the infinite system defined simply by the @xmath4 particles , with @xmath46 say , in some interval @xmath47 $ ] at a given instant in time , substituting @xmath48 in ( [ 1d - exact - force - disp ] ) , and noting that the number of particles in the interval to the left ( right ) of the @xmath35th particle is given by @xmath49 ( @xmath50 ) , we obtain that the force on the particle @xmath35 may be written @xmath51 + 2gn_0[x_i - \langle x \rangle_{\cal
s } ] - 2gn_0 [ \langle u \rangle - \langle u \rangle_{\cal s } ] \label{1d - force - subsystem}\ ] ] where @xmath52 is simply the position of the centre of mass of the subsystem , and @xmath53 the average displacement of the particles in the subsystem
. if the subsystem chosen is the periodic cell of an infinite periodic system , we have @xmath54 and thus obtain ( [ 1d - force - periodic ] ) . for any generic subsystem @xmath45 , on the other hand , the last term on the right in ( [ 1d - force - subsystem ] ) is the same for all particles in the subsystem . defining @xmath55 we thus obtain the equation of motion for any particle in the subsystem , relative to the cm of the subsystem , as @xmath56 where the sum extends over the particles in the subsystem .
this remains valid for as long as no other particle outside the system crosses the particles at its extremities .
( [ 1deom - finite - finale ] ) are thus exactly the equations of motion of a subsystem as long as it is isolated in this sense . for @xmath57 we can transform to the coordinates @xmath58 in which eq .
( [ 1deom - finite - finale ] ) becomes @xmath59r_i\ , .
\label{1deom - finite - phys - coords}\ ] ] these coordinates have been chosen in order to remove the damping term in the equations of motion , and all dependence on the expansion now appears only in a time dependent force term , of which the amplitude decays as @xmath60 . at sufficiently long times , therefore , an isolated subsystem of the infinite 1d expanding " model has , to an arbitarily good approximation , the equations of motion of a strictly isolated non - expanding 1d self - gravitating system .
we will refer to the coordinates in ( [ physical coordinates ] ) as _ quasi - physical coordinates _ : they are the closest analogy we have to what are called physical coordinates in the three dimensional problem ( i.e. @xmath61 , @xmath62 ) . in these
coordinates the 3d cosmological equations of motion are equivalent to those for an infinite newtonian self - gravitating system expanding about a centre in an infinite ( static ) space .
correspondingly , any subsystem ( e.g. galaxy ) taken far from other mass in an expanding universe will evolve in these coordinates like an isolated newtonian system ( see e.g. @xcite for a derivation ) .
the additional term in @xmath60 in the 1d model , which we can not get rid of by a coordinate transformation , appears because we have imposed in the derivation of the model the 3d expansion law rather than using the analogy of this law in 1d ( which gives a collapse in a finite time , qualitatively different to that of the 3d expansion which we wish to model ) . it is important to note that , re - expressed in terms of the three dimensional scale factor @xmath14 , the transformation in ( [ physical coordinates ] ) from the `` comoving '' coordinates @xmath63 to the `` quasi - physical '' coordinates in ( [ physical coordinates ] ) is @xmath64 , and _ not _ @xmath65 as one might naively expect . as a result
they arrive at the incorrect conclusion from their numerical study that stable clustering is not observed in the model . ] . the second term on the right hand side of eq .
( [ 1deom - finite - finale ] ) may be rewritten simply as @xmath66 g n_0 y_i$ ] the first term is of order @xmath67 where @xmath68 is the average comoving density of the subsystem .
thus if we consider any isolated subsystem which is already highly non - linear ( with @xmath69 ) we are necessarily in the regime where the second term may be neglected . in this case
the isolated subsystem will thus behave , in quasi - physical coordinates , like a truly isolated 1d self - gravitating system . just as in 3d , such a system , for large @xmath4 , in the collisionless limit , attains on a few dynamical time scales ( @xmath70 ) a stationary virialized state ( see e.g. @xcite and references therein ) , i.e. , a system which becomes time independent in the quasi - physical coordinates @xmath71 .
more generally , even if not strictly virialized , it rapidly attains a well defined average size about which virial oscillations persist . in order to derive the exponent in what follows we will also use the self - similar behaviour of the evolution , which , for the two point correlation function @xmath72 means that @xmath73 in the spatial range where
the self - similarity is observed .
denotes the reduced two point correlation function , i.e. , @xmath74 where @xmath75 is the local particle density and @xmath76 denotes the average over the statistically translationally invariant point process .
see e.g. @xcite or @xcite . ]
the function @xmath77 is derived from linear theory , which describes a self - similar evolution from power law initial conditions once the growing mode dominates . in the 1d model
the linear theory growth of perturbations can be easily derived by using the expression for the force eq .
( [ 1d - exact - force - disp ] ) in terms of the displacements @xmath34 .
setting , without loss of generality , the mean displacement @xmath40 to zero , we have simply @xmath78 which is characterized by the growing mode @xmath79\ , .
\label{linear theory - growing}\ ] ] for @xmath80 we thus recover the same growth law as in the standard 3d eds model . assuming a power law power spectrum of density fluctuations @xmath81 ( and thus @xmath82 at large @xmath83 ) the evolution is self - similar with @xmath84 shown in fig .
[ fig1 ] is a typical two point correlation @xmath85 obtained from the evolution of power law initial conditions in the class of models we study . between a lower cut - off scale @xmath86 , and an upper cut - off scale @xmath87 a simple power law behaviour of the correlation function
is observed @xmath88 .
the exponent @xmath0 may be written as @xmath89 the temporal evolution of the correlation function is found to be self - similar , down to the scale @xmath90 , below which it is broken .
the scale @xmath91 is in the range in which self - similarity applies , and thus scales in proportion to @xmath92 , with @xmath93 a fixed amplitude corresponding to the transition between the linear and non - linear regime . if we make now the hypothesis that the structures giving rise to the correlation measured at the lower cut - off are _ stable in quasi - physical coordinates _ , i.e. , these structures behave like isolated approximately virialized structures , we have that @xmath94 .
further , since for @xmath95 , @xmath85 mesures simply the average density at distance @xmath83 from an occupied point normalized to the mean density , we have that @xmath96 .
it follows that the exponent tends asymptotically ( at sufficiently long times ) to @xmath97 this result is identical to that derived in @xcite using more heuristic arguments based on the scaling of the energy , and thus @xmath98 . ] .
it tells us how the exponent , in the stable clustering hypothesis , depends on the initial conditions , characterized by the exponent @xmath1 of the power spectrum of density fluctuations , and on the expansion rate parametrized by the constant @xmath2 , which we recall is @xmath99 note that as @xmath100 we obtain @xmath101 , i.e. a correlation function which is predicted to be flat over an arbitrarily large distance . indeed , in this static limit , the stable clustering hypothesis corresponds to stability also in comoving coordinates , and therefore to a constant @xmath86 and @xmath102 .
clearly , however , we do not expect in this case that the formation of non - linear structure at a given scale will leave approximately unperturbed the structures formed previously at smaller scales .
the numerical study in @xcite , as most previous studies of the three models belonging to the class we are considering , exploits fully the very attractive property of these models that they admit exact " numerical integration : the force on particles , whether given in the form ( [ 1d - exact - force - disp ] ) or ( [ 1d - force - periodic ] ) , is such that the motion ( with or without damping ) admits an analytical solution other than at particle crossings .
an event driven algorithm can then be employed , in which the determination of the time of the next collision " , and which pair of particles it involves , requires only the solution of algebraic equations .
as described by @xcite the integration can be sped up optimally using a heap " structure . for the previously considered cases ,
corresponding to @xmath103 , the algebraic equations involved are , respectively , quadratic , quintic and cubic . for a generic @xmath2 ,
however , the equations are not polynomial , and the numerical cost of solving them increases and makes the code , which is already expensive numerically , even more time consuming where @xmath104 , @xmath105 , and @xmath106 and @xmath107 are constants . ] .
as we wish to study fully a broad range of @xmath2 and initial conditions , we have chosen here instead to use a particle - mesh ( pm ) code of which the numerical speed can be greatly enhanced by the use of fft techniques and parallelization . in order to be sure that we are not , as a result , losing the advantages of the precision of the integration accessible in these 1d models , we calibrate , as discussed below , our choice of the new discreteness parameters introduced by the pm code by comparing our results with those obtained , for @xmath103 , using the exact " code .
our pm code defines a regular grid on the simulation box , on which a density field @xmath75 is defined at each time step by a
clouds in cell " interpolation of the particle positions .
the modified 1d poisson equation @xmath108\ ] ] is then solved for the potential in fourier space using an fft , using the solver fftw3 .
an inverse fft then determines the force on the grid , which is then interpolated back onto the particle positions .
the latter are then advanced using a leapfrog algorithm , with a simple adaptative timestep : it is chosen at each time by imposing that the mean distance travelled by particles during a time step be equal to a chosen fraction @xmath109 of the pm grid .
compared to the exact code , our pm code therefore introduces resolution effects controlled by two dimensionless parameters : the size of the pm grid compared to the lattice spacing in the initial conditions , which we denote by @xmath110 , and the parameter @xmath109 controlling the size of the time steps .
we have made an extensive study of the effects introduced by this finite resolution of the pm code , and in particular on the determination of the two point correlation function which is the quantity which interests us here .
shown in fig .
[ fig2 ] is , for example , a comparison of the results for runs from identical initial conditions , for the case @xmath111 and @xmath80 , obtained from the exact code and the pm code , for two different values of @xmath110 , and @xmath109 sufficiently small ( of order unity ) so that convergence is obtained with respect to it .
we find , very reasonably , that excellent agreement is obtained provided @xmath110 is chosen smaller than @xmath86 , while a larger @xmath110 than this minimal scale leads to a visible deviation of the correlation function from its correct value at scales below @xmath110 .
we have chosen our numerical parameters in the simulations reported below in order to obtain such an indiscernible difference in the correlation function between our code and the exact code , for the cases @xmath103 .
all the results presented here are for systems with @xmath112 particles ( in periodic boundary conditions ) . for initial conditions with power law power spectra @xmath113 ( at small @xmath114 , a cut - off at large @xmath114
is always implicitly assumed ) , we expect hierarchical structure formation in @xmath115 dimensions for @xmath1 in the range @xmath116 @xcite : for @xmath117 the variance of mass fluctuations diverges at large scales , while for @xmath118 linear theory breaks down ( and the dynamics at large scales will be determined by the power at the ultra - violet scale ) .
we thus consider ( as in @xcite ) the cases @xmath119 .
point processes with such power spectra are easy to set up : @xmath120 is obtained by randomly distributing points in the interval ( giving a power spectrum @xmath121 ) , the case @xmath111 by applying small random and uncorrelated displacements to a regular lattice , and @xmath122 in the same way but with the additional constraint that neighbouring pairs have equal and opposite displacements ( see @xcite and references therein ) .
our results are summarized in fig .
each point on the plot corresponds to a single simulation with given @xmath1 and @xmath2 , where the latter varies in the range @xmath123 @xmath124 .
the figure shows then the value of the exponent @xmath0 of the power law region of the correlation function . in each case
the given value is determined using a linear regression in the range between @xmath86 and @xmath125 , the latter being determined themselves by eye ( cf .
[ fig1 ] ) .
the uncertainty in the fitted exponent associated with the ( by eye ) estimation of the end - points for the fit is very small ( of order a few percent at most ) in most parts of the parameter space .
however , as we will now discuss and explain , at lower values of @xmath2 and @xmath1 , where the extent of the power law region is much smaller , the resultant uncertainty becomes more significant .
the continuous curves plotted correspond to the predictions of the stable clustering hypothesis , eq .
( [ sc - prediction ] ) .
our results show excellent agreement with the stable clustering prediction for a very large part of the explored parameter space of initial conditions and cosmology .
this region can be well characterized simply by the condition @xmath126 , i.e. , good agreement with stable clustering is observed simply in the region where the predicted exponent be larger than some critical value . in the rest of the parameter space ( i.e. for @xmath127 )
there is clear disagreement with the stable clustering prediction , and the measured exponents lie in a very narrow range , between @xmath128 and @xmath129 , or a little smaller .
it is precisely in this part of the parameter space , however , that there is also a considerable scatter about the curves of the measured exponents .
this scatter is in fact just a measure of the uncertainty in the determination of these exponents , which is most difficult ( for reasons we explain in detail below ) in the region where the exponents become small .
thus , our results are quite consistent with the hypothesis that the exponent in the region @xmath127 is universal " , in the sense that it is independent of initial conditions and cosmology . a schematic representation of this result is shown in fig .
[ fig4 ] : the parameter space of initial conditions and cosmology breaks up into a stable clustering " region , and a region of universality " . the contour defining the two disjoint regions is simply given by @xmath130 . rather than a line
, there may of course be a small region in parameter space in which the exponent varies in a continuous manner from its stable clustering value to a limiting universal value of @xmath131 .
we have found numerically that the stable clustering hypothesis predicts the exponents of non - linear clustering very accurately at sufficiently large @xmath1 and @xmath2 , i.e. , if the spectrum of initial fluctuations is sufficiently blue " and the expansion rate is sufficiently fast .
more precisely it appears that the criterion for it to work is that @xmath132 , which is a monotonically increasing function of both @xmath1 and @xmath2 , be larger than some critical value .
it is not difficult to give a simple physical explanation for this behaviour , as we now explain .
let us consider two overdense regions , initially ( in the linear regime ) of comoving size @xmath133 and @xmath134 .
self - similarity of their evolution implies that there is simply a time delay @xmath135 between every stage of their evolution given by @xmath136 . using eq .
( [ ss - scaling ] ) we obtain @xmath137 let us suppose that the fluctuation @xmath138 goes non - linear and approximately virializes at some given time @xmath139 , when its size is @xmath140 .
when fluctuation @xmath141 virializes , at time @xmath142 , its size is @xmath143 .
if in the interval between @xmath139 and @xmath144 , structure @xmath138 behaves to a very good approximation as an isolated structure , its size in comoving coordinates will decrease by a factor @xmath145 .
it follows that the relative size of the two structures at the time when @xmath141 virializes is @xmath146 where @xmath147 is the growing mode exponent given in ( [ linear theory - growing ] ) . quite simply , the faster the expansion rate , or the larger is @xmath1 , the more a given scale which has entered the non - linear regime can shrink " relative to a larger scale in the time before the latter goes non - linear .
further we note , using ( [ linear theory - growing ] ) , that eq .
( [ scaling ] ) can be written @xmath148 i.e. , the dependence on @xmath3 is in fact completely determined through the exponent @xmath132 itself . the predicted stable clustering exponent is therefore directly related to a physical quantity which we would expect naturally to control the validity of the assumption made in deriving it : the more concentrated " are the pre - existing virialized substructures inside a larger structure when it collapses , the better should become the approximation that this structure will behave simply as a collection of sub - structures which are not disrupted by the virialization and subsequent evolution of the larger structure . indeed , in the limit that @xmath132 approaches unity , any structure which collapses and virializes will see the substructures it contains essentially as point particles . as @xmath132 decreases , on the other hand , we expect that the interaction between structures can lead to their disruption , and in particular that mergers " of substructures become much more probable .
given that the value of the exponent @xmath132 characterizes precisely the relative condensation " of scales , it is very natural that a critical value of the exponent should characterize the breakdown of stable clustering .
what our results do not allow us to conclude , as we have discussed , is what precisely happens in the regime where stable clustering breaks down : our numerical results do not allow us to clearly distinguish between the possibility of an abrupt ( discontinuous ) transition to a region in which there is a truly universal exponent , or a smoother transition to a region in which the exponent depends only very weakly on the relevant parameters ( @xmath149 ) .
further we do not ( currently ) have a model to explain the exponent ( or narrow range of exponents ) observed in this part of the parameter space . extending the analysis given just above to determine the scalings of relative sizes of structures , it is simple to understand why it is precisely the region of small @xmath132 in which the numerical results for the exponent of the power law in the correlation function are most noisy .
the degree of precision in the measured exponent is essentially just a function of the range of scale over which the power - law behaviour extends , i.e. , it depends on the ratio @xmath150 accessible in a numerical simulation .
the scale @xmath87 corresponds approximately at any time to the size of the largest ( approximately ) virialized scale .
let use denote by @xmath6 the comoving size of the _ first _ structure which virializes in the simulation , at the time @xmath151 at which it virializes .
self - similarity and stable clustering then imply the temporal evolutions @xmath152 we can infer , using eq .
( [ time - delay ] ) , that , the range over which power law scaling is expected for @xmath153 at the end of the simulation can be expressed as @xmath154 where @xmath155 and @xmath156 are , respectively , the initial comoving size of the first and last scale to virialize during the evolution , and @xmath157 ( @xmath158 ) are the number of particles they contain , respectively .
thus , while for @xmath132 closer to unity this range of scales is large , as @xmath132 decreases it contracts and is simply of order the ratio @xmath159 for @xmath160 . for our simulations , with @xmath112 particles , taking @xmath161 , or a little larger , and @xmath162 , we can , for small @xmath132 , access power law clustering over approximately two orders of magnitude , which leads in practice to the observed order of the uncertainly in the exponent ( @xmath163 ) .
let us compare our results to those in three dimensions .
numerical simulations have been done by various groups to test the stable clustering hypothesis , via the study of the eds model starting from power - law initial conditions ( see , e.g. @xcite ) . in this case , corresponding to @xmath80 in our notation , the predicted exponent , in three dimensions , of the two point correlation function is @xmath164 .
the range of power spectra which has been probed numerically is @xmath165 .
further , more generally , the stable clustering hypothesis leads to scaling relations for higher order correlation functions , which have been tested by some authors @xcite . until the most recent and extensive study of @xcite , previous works had concluded that the stable clustering prediction worked quite well , with , in some cases , indications of small deviations close to the limits of numerical resolution .
this last study , on the other hand , reports measurements of the exponent @xmath0 which are clearly discrepant with the stable clustering prediction : for @xmath166 , @xmath167 rather than @xmath168 ; for @xmath169 , @xmath170 rather than @xmath171 ; for @xmath172 , @xmath173 rather than @xmath174 ; for @xmath120 , @xmath175 rather than @xmath176 .
these results are clearly _ qualitatively _ different from those we have found in our 1d models , for any @xmath2 . while the measured @xmath0 increases as @xmath1 increases , just as in one dimension , the discrepancy with stable clustering in three dimensions is associated with an exponent @xmath0 which is _ smaller _ than the predicted stable clustering exponent .
further there is no indication in this limited range of @xmath1 probed that the discrepancy with respect to stable clustering is increasing , nor that the associated discrepancy manifests itself as an ( exact or approximate ) universality of the measured exponent .
these differences between 1d and 3d results could , of course , be simply a reflection of the difference of the gravitational clustering in the two cases .
one difference which may be essential , as discussed in @xcite , is that in one dimension there are no tidal forces , and a virialized structure may only be perturbed by another structure actually crossing it ( which is , on the other hand , much more likely than in three dimensions ) .
however , one of the main motivations for this study is precisely that the reliability of results from 3d simulations is not clear : the combination of the fact that the range of comoving scales probed grows in proportion to @xmath177 , and the necessity to introduce a ( relatively large ) smoothing parameter to regularize the small scale divergence in the 3d gravitational force , mean that the exponents ( in existing studies ) are measured over at most a little more than one order of magnitude . further there are consideable discrepancies between some of the studies .
@xcite find , for example , good agreement with stable clustering for the case @xmath166 with simulations of a similar size to those of @xcite .
let us focus a little more on these differences between 1d and 3d simulations linked to the small scale smoothing of the force .
in 1d , as discussed , the equations of motion can be integrated exactly in absence of smoothing , so that there is in practice _ no lower limit _ on spatial resolution other than that arising from the finite particle density , of order the scale @xmath86 in the correlation function .
we note that , following the analysis above , we infer that , if stable clustering holds , we have @xmath178 in one dimension , while the same arguments applied to three dimensions give @xmath179 the scale @xmath180 , the comoving size of the first structure which virializes , is of order the initial interparticle spacing @xmath181 . in one dimension , even when we use , as here , a pm code , the smoothing can , without excessive numerical cost , be chosen so that it is at all times considerably smaller than the scale @xmath86 , so that we can be absolutely confident that it plays no role in the evolution . in three dimensions , on the contrary , cosmological simulations of a reasonable size ( i.e. , so that the range of scales @xmath159 which go non - linear cover at least a decade or two ) are typically performed with a smoothing of between one tenth and one hundredth of the interparticle spacing @xmath181 . thus 3d simulations can not in practice simulate systems manifesting stable clustering only over a very limited range of length scales , while remaining in the regime where @xmath182 . while the use of a cut - off @xmath183 does not necessarily imply that the clustering above the scale @xmath184 is not accurately reproduced , it is quite possible that this could be the case if great care is not taken in the choice of numerical parameters ( see @xcite ) . in this respect
even when a relatively large smoothing is employed , two body relaxation may play a role in disrupting the desired collisionless evolution of @xmath4-body simulations , and such effects can potentially lead to a pollution " of larger scales than @xmath184 @xcite . in the 1d system , in contrast , there is in fact no two body collisionality analagous to that in three dimensions , and collisional relaxation ( of which the mechanism is not yet fully understood ) is extremely slow compared to that induced by 3d two body relaxation(see @xcite and references therein ) .
simple estimates show that such relaxation should play absolutely no role in our 1d simulations , while the same is not true in typical 3d simulations .
our study thus suggests that particular attention should be paid to this point in the analysis of 3d simulations .
we note also that simulations of clustering in a 3d universe have been performed for the case of a static universe , for the cases @xmath120 @xcite and @xmath111 @xcite .
the results are in this case qualitatively very consistent with those observed here and in @xcite for the static ( @xmath185 ) limit . just as in the 1d simulations self - similarity is observed , and further a correlation function which appears to be the independent of the initial conditions , with a very shallow exponent in the inner part , @xmath186 .
finally let us discuss the question of universality " .
a striking feature of our 1d results is that the measured exponent characterising clustering appears to become independent of initial conditions and cosmology when the stable clustering approximation breaks down more specifically when the exponent @xmath1 in the initial power spectrum is sufficiently small for any given expansion rate parameter @xmath2 . in three dimensions , following notably the work of @xcite ,
much emphasis has been placed on a universality " of cosmological gravitational clustering which manifests itself as an apparent independence of halo profiles of initial conditions and cosmology .
the range over which such a universality might apply is not , however , clear : studies such as that of @xcite show that , even in the eds cosmology , the inner exponent of halos may vary sensibly for power law initial conditions as a function of @xmath1 . does the apparent universality we observe in 1d correspond to a universality in halo profiles ? as discussed in @xcite , one can identify and extract halos by prescriptions analogous to those used in three dimensions .
however , the objects thus extracted are simply so clumpy " , at any scale significantly above the scale @xmath86 , that it is not possible to approximate the run of density by a smooth profile : indeed it can be verified directly that the power law exponent ( over several decades ) in the correlation function corresponds to a truly scale invariant fractal distribution of the matter in the corresponding range .
if , however , the extracted halos are smoothed ( by hand ) over a scale much larger than @xmath86 , the exponent characterising the decay of density about the centre of the halo , should be approximately the same as that measured in the correlation function .
this is true because the latter measures , by definition , the mean density as a function of distance about any point , which in a simple fractal type distribution is ( modulo fluctuations ) the same about any point .
thus , in the 1d model , the universality we observe here is indeed expected to be associated to a universality of inner exponents of halos . extrapolated to 3d , our results thus suggest that the relative smoothness of halo profiles is an artefact of insufficient spatial resolution , and that the minimal scale of substructure is limited only by such resolution . in this case , for power law initial conditions , the observed power law slope of the correlation function should in fact _ coincide _ with the inner exponent of the measured halo profiles .
we note that this is completely different to what is predicted usually in three dimensions on the basis of a smooth halo model ( see e.g. @xcite ) . up to now 3d numerical studies have not , to our knowledge , addressed this point directly ( by comparing measured slopes in halos and two point correlation functions , in scale free cosmologies ) .
we note that the results of @xcite on inner halo profiles appear , however , to be very consistent with a behaviour completely analogous to what we have seen in our 1d models : in their study of power law initial conditions in an eds cosmology , for @xmath1 varying in a range between @xmath187 and @xmath188 , inner slopes for halo profiles are estimated using different fitting procedures . for the simplest such procedure it is found ( see fig .
2 ) that the measured exponent varies from a value in good agreement with stable clustering for the largest value of @xmath1 , but appears to deviate , as @xmath1 decreases , towards an exponent which is _ larger _ than predicted by stable clustering , tending rapidly towards an asymptotic exponent a little larger than unity , and consistent with that of @xcite .
this suggests that in the 3d eds model , the universality of halos would translate to an insensitivity to initial conditions below an @xmath1 of order @xmath189 .
given that the range of effective exponents of initial conditions of typical cosmological ( e.g. @xmath25cdm ) initial conditions falls in or close to this range , this would appear to be very consistent with the approximate universality of halo profile exponents observed in these kind of simulations .
our 1d study thus motivates further careful numerical study of non - linear clustering in the pure matter ( @xmath80 ) eds model in three dimensions , extending recent studies such as @xcite and @xcite . in the same way
as has been done here , the study of this model can be extended to the full family of eds models of which one limit is a static universe .
the goal of such a study would be to see if , in this space of models , and for power law initial conditions , a similar structure is found to that we have seen in one dimension .
our 1d study suggests that particular attention needs to be applied to controlling for the robustness of measure of exponents in the two point correlation function to small scale force smoothing , particularly when an exponent below the predicted stable clustering value is found ( as in @xcite ) .
further , study of the relation between exponents measured from the two point correlation function and those measured from halos , and their variation as a function of @xmath1 and @xmath2 , should clarify whether the qualitative nature of clustering is indeed like that usually assumed in theoretical modelling ( smooth halo models ) or instead like that in the 1d models ( hierarchical fractal clustering ) .
if the framework suggested by the 1d models of a universal " region in the space of initial conditions of clustering , delimited by a critical " value of the predicted stable clustering exponent is correct , and applies also in three dimensions , the theoretical problem remains of understanding both the corresponding value of the universal exponent " , and why this universality applies in the specific region where it does .
further study of the 1d model may then help to throw further light on this . as the static model is in the relevant part of the parameter space
, it may suffice to study this particular limit , i.e. , the region in which there is universality is where the expansion of the universe has little effect on the non - linear clustering .
we note that a very recent study in @xcite shows that there is an apparent universality in the properties of the halos " formed from the 1d collapse of a single structure with a range of cold initial conditions , and argues that this universality may be related to that observed numerically in three dimensions . the 1d halos in @xcite , however , are smooth , very different to the non - linear structures we observe here , which are very clumpy even in the static limit of the 1d model .
as discussed above , however , in this limit the range of non - linear clustering explored by our simulations is still quite modest , and it may be that larger simulations might establish a link to the study of @xcite . with respect to this last point
the effect of varying particle number we note that we have not discussed here the question of whether the dynamics of these simulations in the non - linear regime is representative of the fluid or vlasov poisson limit . in three dimensions
this is a crucial question which has been the subject of discussion and some controversy @xcite , and one might expect that these 1d models , with their large accessible spatial resolution , could help to throw light on this question . in n body simulations this question can in principle be directly probed by comparing simulations in which the particle number n is varied while keeping the lower cut - off scale to the initial density fluctuations fixed , or , alternatively , by increasing particle density in units of the grid of our pm code .
further , in contrast to three dimensions , it should be feasible to address this question much more directly by comparison of the results of the integration of the n body system with a direct numerical integration of the vlasov - poisson equations themselves .
this is another interesting future direction for work on these models .
we are indebted to b. marcos for numerous discussions and suggestions .
we also thank s. colombi , b. miller , j. morand , j .- l .
rouet , f. sylos labini , p. viot and t. worrakitpoonpon for useful conversations . | one dimensional versions of dissipationless cosmological n - body simulations have been shown to share many qualitative behaviours of the three dimensional problem .
their interest lies in the fact that they can resolve a much greater range of time and length scales , and admit exact numerical integration .
we use such models here to study how non - linear clustering depends on initial conditions and cosmology .
more specifically , we consider a family of models which , like the three dimensional eds model , lead for power - law initial conditions to self - similar clustering characterized in the strongly non - linear regime by power - law behaviour of the two point correlation function .
we study how the corresponding exponent @xmath0 depends on the initial conditions , characterized by the exponent @xmath1 of the power spectrum of initial fluctuations , and on a single parameter @xmath2 controlling the rate of expansion .
the space of initial conditions / cosmology divides very clearly into two parts : ( 1 ) a region in which @xmath0 depends strongly on both @xmath1 and @xmath2 and where it agrees very well with a simple generalisation of the so - called stable clustering hypothesis in three dimensions , and ( 2 ) a region in which @xmath0 is more or less independent of both the spectrum and the expansion of the universe .
the boundary in @xmath3 space dividing the stable clustering " region from the universal " region is very well approximated by a critical " value of the predicted stable clustering exponent itself .
we explain how this division of the @xmath3 space can be understood as a simple physical criterion which might indeed be expected to control the validity of the stable clustering hypothesis .
we compare and contrast our findings to results in three dimensions , and discuss in particular the light they may throw on the question of universality " of non - linear clustering in this context .
cosmological structure formation , gravitational clustering , @xmath4-body simulation |
the large absorption line red shift measurements settled the issue that most @xmath0-ray bursts are at cosmological distances .
the physical mechanism of the origin of the @xmath0-ray bursts , questions of whether there are more than one mechanism , and the issue of whether @xmath0-ray bursts interact only with the interstellar medium or with a circumstellar environment are still very open . correlations with star formation and observations of putative iron lines with the chandra observatory @xcite have put new focus on the connection to a supernova - like core collapse event . currently , most observations of @xmath0-ray burst afterglows are photometric , with the effort being to find the power law behavior of the afterglow brightness as a function of wavelength and time .
most of the 25 known red shifts have been measured by deep spectroscopy on the host galaxy long after the afterglow has faded .
prior to grb 021004 , only eight @xmath0-ray burst afterglows have had early - time spectra that reveal absorption lines from the host galaxy ( e.g. , vreeswijk et al .
2001 ) . of these , virtually all have been low excitation lines in one velocity system that are likely to arise in gas far from the burster .
two exceptions to this are grb 011211 which shows and in addition to lower excitation lines @xcite , and grb 010222 which displayed two low - ionization systems separated by @xmath6 km s@xmath1 that could arise from clouds at different positions within the host galaxy @xcite .
grb 020813 has a spectrum which displays absorption lines in both a z=1.223 and z=1.255 systems @xcite .
with so little known , and many afterglows revealing new properties , more prompt spectroscopy is clearly needed .
afterglow spectra could show absorption lines arising from the host galaxy ( especially the fe and mg ism lines ) that will give the red shift .
the red shift is the key parameter for determining the burst energetics and hence further testing the conclusions of @xcite and @xcite that @xmath0-ray bursts have a nearly constant total energy .
one can also improve the calibration ( currently based on 9 events , see also schaefer , deng , & band 2000 ) that the spectral lags and light curve variability are cepheid - like luminosity indicators that could turn @xmath0-ray burstsinto a premier tool for cosmology out to z@xmath410 @xcite .
intrinsic afterglow spectral features of any kind would be a boon to interpretation .
if there are underlying supernova - like explosions , corresponding spectral features could be superposed on the continuum and revealed by careful subtraction of the well - defined power - law continuum .
detection of , for instance , features of red shifted ca h & k , which tend to be among the strongest lines in all types of supernovae , would yield both a red shift and clues to the nature of the explosion itself .
other possibilities would be to discover fe emission lines , which would yield temperatures , red shifts , and abundances .
perhaps the most important possibility is that of serendipitous discovery of some unexpected feature .
grb 021004 provided an unprecedented opportunity to obtain prompt spectroscopy of the early afterglow of a @xmath0-ray burst .
grb 021004 was discovered by instruments on the hete ii satellite at 12:06 ut on 2002 october 4 @xcite .
prompt discovery of the optical transient by @xcite allowed rapid spectroscopic observations .
@xcite obtained the first spectra and identified two intervening systems at z = 1.38 and 1.60 from and absorption .
@xcite confirmed those features , noted several absorptions at these redshifts , and pointed out four absorption lines at @xmath4 4633 , 4664 , 5109 , and 5152 ( see also sahu et al .
2002 ; castander et al .
these were identified by @xcite as and features at red shift of @xmath7 .
chornock & filippenko also identified ly@xmath8 emission at z = 2.323 and absorption components and perhaps ly@xmath9 at similar red shifts .
@xcite identified four absorption components of at z = 2.295 , 2.298 , 2.230 and 2.237 , noting that the total spread is about 3000 km s@xmath1 .
they remarked that this velocity dispersion would be difficult to interpret as due to a single cluster of galaxies and noted that if associated with a supernova shell the supernova must have substantially preceeded the @xmath0-ray burst to be optically thin .
@xcite suggested the identification of one absorption feature as low excition at z = 2.328 and identified that , and the ly@xmath8 emission , as the red shift of the host galaxy .
they raised the issue of the velocity dispersion being due to a wind in a wolf - rayet like progenitor .
@xcite noted that the ly@xmath8 flux may imply a strong star formation rate in the host galaxy .
@xcite reported a very high resolution spectrum ( r = 50,000 ) that revealed the complex fine structure of many of the high ionization lines .
they determined that the individual features show complex flow spanning about 1000 km s@xmath1 and conclude that the @xmath10 system is probably close to the @xmath0-ray burst .
they note that of the four high red shift systems , the low ionization lines , and possibly , only occur in the highest red shift , @xmath10 , component . @xcite and mller et al .
( 2002 ) have provided an excellent compendium of line identifications along with detailed line profiles .
@xcite give a spectral index in the optical of @xmath11 , and report a reddening of the continuum in the blue region of the spectrum over the first three days .
here we report spectra obtained within the first day after outburst and five days later .
we concentrate on constraining the circumburst medium in the context of the high ionization lines and the afterglow light curve .
spectra of the optical afterglow were obtained with both the 2.7 m harlan j. smith telescope using the igi imaging spectrograph @xcite and the 9.2 m hobby - eberly telescope ( het ) with the marcario low - resolution spectrograph @xcite at mcdonald observatory .
three consecutive exposures of 900 seconds each were taken with the 2.7 m telescope beginning 14.31 hours after onset of the @xmath0-ray burst , followed by sets of three 900 second exposures taken with the het beginning at 15.48 hours , 20.31 hours , and 4.84 days after the burst .
a 2.5 arcsecond slit was used for the 2.7 m data , while the het observations utilized a 2.0 arcsecond slit , giving all data sets a 16 resolution .
the spectra were reduced using iraf , and wavelength calibrated using argon , neon , cadmium , and mercury lamps . the combined spectrum from the first night
is displayed in figure 1 .
our first night spectrum shows a collection of absorption lines longward of 5800 that arise from iron and magnesium of relatively low ionization at red shifts of 1.38 and 1.60 .
these are due to intervening galaxies with no relevance to the @xmath0-ray burst analysis .
the lines at red shifts of @xmath12 are presented with their measured properties in table 1 .
the lines are separated into two primary velocity components , with @xmath13 and @xmath14 .
we identify the 4587.2 line as being from ( c.f .
matheson et al . 2002 ) .
there are few resonance lines from the ground state of highly - ionized elements of high cosmic abundance in the observed wavelength range , and the only one that has not produced a detected absorption line is with rest wavelengths 1238.8 and 1242.8 for which we can only present an upper limit to its equivalent width .
the separation of the two high redshift primary velocity components corresponds to a relative velocity of 3000 km s@xmath1 in the frame of the host galaxy .
the unblended lines for the @xmath10 velocity component are significantly resolved in our spectrum , with a fwhm of @xmath15 km s@xmath1 . from our own data , this implies either multiple components or one component with a velocity range that large .
other spectra @xcite have resolved the high red shift component into at least two sub - components with a velocity separation of 560 km s@xmath1 .
thus , the @xmath16 system consists of at least three separate velocity components , with velocities of 0 km s@xmath1 , 560 km s@xmath1 , and 3000 km s@xmath1towards earth relative to the highest red shift component .
@xcite presented detailed calculations of the complex of absorption lines produced from 1110 - 1705 by vibrationally excited molecular hydrogen in a molecular cloud near a @xmath0-ray burst .
individual lines have a typical depth of 20% for normal low - resolution sprecta and might be difficult to detect without very high signal . fortunately , the entire pattern of lines is characteristic of the process and so all of the lines ( with their unique positions and depths ) can be simultaneously sought in any spectrum .
the idea is that a spectrum in which any one line would be highly insignificant might yet have the presence of many lines detectable at a high confidence level .
even a noisy spectrum might have its apparent noise following the characteristic molecular hydrogen features hence resulting in a significant detection .
many of the individual lines are extremely saturated with optical depths of @xmath17 in the line center and with equivalent widths of 0.1 or greater , so that the strength of the absorption is quite insensitive to the column density of vibrationally excited molecular hydrogen , @xmath18 .
b. draine has kindly provided us with the transmission spectrum for excited molecular hydrogen for our spectral resolution .
we have used this as a template to cross correlate with our spectrum . for a red shift of 2.3
, we cover the range from 1240 - 1705 .
we find no significant correlation peak for any red shift of the template . within the red shift range of 2.292.33 ,
the largest peak in the cross correlation is never more than the rms of the cross correlation over a much wider red shift range .
the implication of this negative result will be discussed in section 3.5 .
we found no significant differences among our spectra within the first night .
specifically , the equivalent widths of the absorption lines were all identical to within the errors .
we have constructed broad - band photometry from our reduced spectra using the photometric pass - bands of landolt ( 1992 ) and normalizing the colors to match the observations concurrent to the 2.7 m spectrum .
the resulting synthetic photometry for the b , v , and r magnitudes yields colors that are constant to within 0.04 mag for the first night . for a comparison between our observations on the first and fifth nights
, we are handicapped by the poor signal on the fifth night due to the faintness of the afterglow
. the only lines for which we can quote line widths on the fifth night are the two lines which have equivalent widths in the rest frame of the host ( the observed width divided by 1+z ) of @xmath19 and @xmath20 at the same wavelengths as for the first night .
this is consistent with the first night .
we confirm the report by @xcite that the blue end of the continuum became redder from the first night to subsequent nights .
in particular , our synthetic photometry shows the b - v color to have reddened by @xmath21 mag from the first to the fifth night , while the v - r color was unchanged , @xmath22 mag , over the same interval .
the lyman alpha _ emission _ line is likely to represent the average velocity of the host galaxy , although its exact wavelength will shift redward somewhat due to the absorption component . the red shift of the ly@xmath8 line is variously given as 2.3351 , 2.332 , 2.328 , and 2.323 @xcite .
the average velocity of the host galaxy is also likely represented by the low - ionization feature of at 2.328 .
the highest red shift components of the and lines correspond to this host galaxy velocity .
the velocity of the burster itself might vary by up to several 100 km s@xmath1from the average velocity for the host .
the other primary components of these high - ionization lines are at velocities of 560 km s@xmath1 and 3000 km s@xmath1blueward of the host galaxy velocity .
the presence of the @xmath12 lines of and shows that the absorbing gas is highly ionized .
such lines are commonly seen in the spectra of distant quasars through intervening galaxies where the ionized material is in a collisionally ionized halo .
could the absorbing gas for grb 021004 just be in halos of chance galaxies along the line of sight ?
the presence of three velocity components within the @xmath12 system with separation of up to 3000 km s@xmath1 argues strongly that this can not be .
the probability of two galaxies lying in so small a red shift range just blueward of the host is rather small even in a cluster of galaxies , and the velocities are too large to allow for a bound cluster .
sargent , steidel , & boksenberg ( 1988 ) present statistics on lines with rest frame equivalent widths of more than 0.3 ( i.e. , a threshold much smaller than the observed lines for grb 021004 ) , and find that absorption systems with @xmath12 occur with a rate of @xmath23 per red shift unit .
the probability of getting two absorbers within a range of 0.033 in @xmath24 in front of the host galaxy is thus 0.0024 .
such an occurrence is not impossible ( c.f . the @xmath25 system for q1511 + 091 , sargent , steidel , & boksenberg 1988 ) , but it is too small to be plausible for grb 021004 .
it is also improbable that random intervening galaxies will have equivalent widths as thick as 1.4 ( as for the observed lines from the systems blueward of the host galaxy ) .
the distribution of equivalent widths ( for both members of the doublet ) falls off exponentially such that the probability for any one galaxy being above this threshold is @xmath26 , or 4% for both systems . in all , the probability of getting two chance galaxies with a red shift within 0.033 of the host and with rest frame equivalent widths of @xmath27 is 0.00010 .
we conclude that the high - ionization absorption lines are not from intervening galaxies .
the only other situation for which the and lines are seen in absorption is when gas is subjected to a high flux of ionizing radiation , as in an agn , a wolf - rayet star , or a @xmath0-ray burst .
an agn at the position of the @xmath0-ray burst would already have been detected , and so this possibility can be rejected . with only @xmath28 wolf - rayet stars in our own galaxy ,
the odds of the line of sight passing close to 13 such stars is small ( ignoring clustering ) provided that the star is not the progenitor of the @xmath0-ray burst . that the progenitor is itself a wolf - rayet star is plausible , because the progenitor is likely to be a massive star just before a core - collapse
alternatively , the ionizing radiation could come from the @xmath0-ray burst itself .
thus , we think that the most likely means to ionize the gas to high levels is either radiation from the burst progenitor or the burst itself , with both cases requiring the ionized gas to be physically associated with the burster .
wolf - rayet stars have temperatures from 30,00070,000 k and supergiant luminosities . for the hotter stars ,
this is sufficient to ionize the surrounding nebulosity to levels where a significant fraction of the carbon and silicon atoms can produce and absorption and emission .
thus , a wolf - rayet - like progenitor might already ionize the surrounding gas such that later light from the afterglow will have silhouettes of these high - excitation lines .
alternatively , the @xmath0-ray burst itself is an obvious source of a huge ionizing flux .
the most frequent mode of ionization of carbon and silicon will be for the @xmath0-ray to knock out an inner electron , whereupon autoionization will cause ejection of the valence electrons .
in addition the ejected electrons will have sufficient energy to ionize other atoms . in the remainder of this section
, we will estimate a crude distance range over which the burst flux can ionize enough gas so as to produce the observed absorption lines .
the cross section for photoelectric interactions varies strongly with photon energy , so the probability that any given atom is ionized by burst flux requires an integral over energy .
the cross sections for carbon and silicon are tabulated in @xcite along with a prescription for interpolation . for carbon ,
the cross sections are 0.0213 , 40.3 , and 44500 barns ( one barn equals @xmath29 ) at photon energies of 100 , 10 , and 1 kev respectively . with this
, we see that most of the ionization will be from burst photons in the x - ray band .
the burst spectrum has only been measured down to a photon energy of 7 kev ( 23 kev in the frame of the host galaxy ) by hete-2 . above 7 kev , @xcite state that the spectrum is well fit by a single power law such that @xmath30 and has a fluence from 7400 kev of @xmath31 .
below 7 kev , the spectrum will likely turn over to asymptote at a slope close to the synchrotron limit of @xmath32 . here
, we will model this behavior as a broken power law with indices of -1.64 and -0.66 above and below a break energy , @xmath33 .
we also impose a lower energy cutoff in our calculations at @xmath34 , below which we assume that no flux reaches the gas .
this is to model the possible photoelectric absorption of photons before they reach the gas .
we normalize this spectrum to the 7400 kev fluence reported by hete-2 and correct the photon energies of this observed spectrum by a factor of @xmath35 to get the spectrum as seen by gas in the host galaxy frame .
the expectation value for the number of times that a given atom will be ionized by burst flux is @xmath36 , where @xmath37 is the burst photon spectrum ( in units of photons @xmath38 ) as a function of distance from the burster , @xmath39 is the cross section for ionization as a function of photon energy @xmath40 , and @xmath41 is the time . for these calculations ,
we adopt @xmath42 kev .
we find that the result is highly sensitive to the value of @xmath34 ( due to the large cross section for ionization at low energies ) , which itself is highly uncertain since we do not know the degree of absorption of the low energy photons before the radiation hits the gas .
for @xmath43 kev in our rest frame ( 1 kev in the host galaxy ) , we expect @xmath44 ionizing interactions per carbon atom at a distance of 1 pc . for @xmath45 kev in our rest frame
( 0.1 kev in the host galaxy ) , we expect @xmath46 ionizing interactions per carbon atom at a distance of 1 pc
. the expected number of ionizing interactions will scale as the inverse square of the distance . for this range of cases , the expected number of ionization interactions is unity at distances ranging from 10 - 25 pc .
this provides an order - of - magnitude estimate of the minimal range out to which the burst flux will completely ionize carbon .
the radius at which significant high ionization absorption can be caused may be substantially farther than this calculated distance for two reasons .
first , only a small fraction of the carbon need be ionized to result in significant absorption .
since the cross section for the resonance line is large , the intervening gas need only have a column density of @xmath47 in the state to account for the observed absorption line .
this corresponds to a fraction of @xmath48 of the total carbon expected in a smooth wind ( see 3.4 ) .
if only 0.01% of the carbon must be ionized to to create the necessary column density , then the burst s effective radius for ionization is larger than the previous distance estimate by a factor of 100 .
this factor depends on the column density of the cloudy ism .
second , the electrons ejected from the neutral atoms at high velocity will also have a high cross section for ionizing further atoms .
this effect depends on the density of the gas and the spectrum of the electrons @xcite .
this effect will roughly increase the ionizing radius by a factor of the square root of the ratio of the electron energy to the ionization energy , which might be up to a factor of order ten . in all , there are many uncertainties and dependencies on unknown conditions , but the effective ionization radius of the burster might be of order 10100 times larger than calculated in the previous paragraph .
this would imply that any gas cloud within perhaps 1002500 pc of the burster will suffer enough ionization to create a sufficient column density of so as to form a detectable absorption line .
the complementary question is how close to the burster some gas will be incompletely ionized at the epoch of observation .
this is important since we later consider the possibility of the absorbing gas being quite close ( @xmath26 pc ) to the burst .
@xcite have presented a detailed and extensive calculation that shows that even a very thick wind will have all of its hydrogen completely ionized by a grb flash on time scales of milliseconds or faster .
the implication is that most ( if not all ) of the atoms of carbon in a wind will also be completely ionized , and thus there may be no atoms in the state to create absorption lines near the burst . whereas lazzati & perna considered a smooth wind , the column density of @xmath49 @xmath50 could be accounted for by clumps in the gas .
the existence of such clumps is plausible since images of wolf - rayet winds show structure on all size scales .
also , the quasi - stationary flocculi ( clumps of gas created in the wolf - rayet progenitor wind ) in cas a and the kepler supernova remnants have densities of @xmath51 @xcite after correcting by a factor of 4 for the shock jump conditions , with a typical projected size of @xmath52 cm in width and @xmath53 cm in length . for grb 021004 in particular , the existence of clumps is indicated by the three bumps in the afterglow light curve ( see 5 ) where the clumps might be large in size with moderate contrast or might be small in size yet with large central densities .
the existence of clumps in the winds of massive stars is thus expected , and these clumps can have densities @xmath54 @xmath55 .
clumps can serve in two ways to allow for the existence of close to the @xmath0-ray burst .
first , the clump will shield portions of the wind material ( both within the clump and within the shadow of the clump ) from the high burst flux . for example
, a clump with size @xmath56 cm with a density of @xmath57 at a distance of a fraction of a parsec from the burster should absorb the ionizing flux .
second , the higher density within clumps can allow for faster recombination that can generate a day after the burst even if the burst had completely ionized the carbon .
the degree of recombination that will have occured when we see the absorption a day or so after the burst has ionized the gas depends strongly on the density . for ionized carbon ,
the recombination time scale is @xmath58 days for densities of @xmath59 that would be typical of a wolf - rayet wind at a distance of @xmath60 pc .
the recombination time scale would be @xmath44 days for densities of @xmath57 that are typical of the densities observed in supernova remnant knots that are presumed to have formed in the progenitor s wind .
within such clumps , the fraction of carbon in the state should be @xmath61 a day after the burst . a clump with density @xmath57 and size @xmath56 cm will produce a column of @xmath62 in .
this is @xmath63 times larger column than is required to produce the observed absorption lines .
detailed calculations of the ionization structure are difficult without knowledge of the density structure in the wind .
nevertheless , it appears that shadows and recombination in the expected clumps can yield a column density of @xmath47 even quite close to the burster .
the calculation of the ionization structure is not straight forward even if the density structure is given .
one trouble is that the burst spectrum below 7 kev is unknown , and this is the critical energy region for ionizing the gas due to the large cross sections for photo - ionization .
a related problem is accounting for the ionizing flux from the afterglow , which is comparable to that from the burst itself .
an additional problem is that stimulated recombination will ensure that some fraction of the carbon will not be completely ionized even in a radiation field of the highest intensity . yet
another issue is that the photo - ionized electrons likely have roughly a power - law energy distribution , whereas the cross sections for ordinary and stimulated recombination are calculated for thermal electron distributions .
the correct recombination rates are likely to be significantly different than the usual " calculations . in principle
, these problems can all be surmounted by substantial theoretical effort . in the mean time , we conclude that it is difficult to make any definite statements regarding the ionization state of gas close to the burster .
a cloud far from the @xmath0-ray burst will have a low relative velocity corresponding to ordinary galactic rotation .
such a cloud would be an obvious candidate for creating the high - excitation low - velocity absorption component .
another obvious candidate to contribute to the high - excitation low - velocity lines is the shell piled up by the o main sequence star wind as it blows a bubble in the local ism at a distance of perhaps 30 pc .
thus , there are two adequate sources of material along the line of sight with distances from 302500 pc that can be ionized by the burst flux .
gas near the burster will receive a tremendous blast of photons , and interactions will transfer some of the photons momentum to the gas .
the total radiative acceleration of the gas that determines its final velocity will occur during the few seconds when the burst flux passes through the gas a few hours before the time at which the absorption lines are produced by the gas . how much momentum gets transferred depends on the cross section of the gas , the spectrum of the burst , and the flux of the burst photons .
the final velocity will fall off as the inverse - square of the distance from the burst . at some distance
, the blast will accelerate the gas to 3000 km s@xmath1 , and this process could account for the observed high relative velocity of one of the absorption components . in this subsection
, we present a detailed calculation of the radiation - induced velocity as a function of distance from the burster .
the radiative acceleration @xmath64 ( as a function of distance r and time t ) by a central energy source is given by @xmath65 where @xmath66 is the cross section for interaction with a burst photon of energy @xmath40 , @xmath67 is the mass per nucleon , and @xmath68 is the eddington flux which is the first moment of the intensity ( with units of erg @xmath69 .
the velocity due to radiative acceleration is @xmath70 where @xmath71 is the initial velocity of the gas , and @xmath72 is the time interval covering the burst duration as the photons pass through the gas being accelerated .
this integral over time will change the burst flux into a fluence , which is to say that the acceleration only depends on the number of incident photons and not on their time distribution .
this change in velocity structure in the gas around the burst will not lead to any significant changes in the density structure on time scales for which we are concerned .
that is , a parcel of gas accelerated to , say , 3000 km s@xmath1 will move only 1.7 au in one day , so as to cause a @xmath73 density change within a gas clump of size , for example , @xmath56 cm if the backside suffers zero acceleration .
the @xmath0-ray cross sections for the photon - matter interaction have been included as detailed in hflich ( 1991 ) .
because the photon energies far exceed the binding energies of the electrons in atoms , the interaction of electrons and photons is given by the klein - nishina cross section per nucleon .
for lower energies , we determine the angle dependent klein - nishina cross section by a standard rejection technique and obtain the cross section by formal integration over angle .
bound - free cross sections have been included according to veigel ( 1973 ) .
the x - ray opacities are dominated by bound - free transitions from inner shell electrons ( which are always populated since when an inner shell electron is ejected it is rapidly filled by one of the outer electrons ) , and are independent of the conditions in the gas cloud .
thus , the bound - free opacity is rather model independent .
the final velocities are a function of composition and the burst spectrum . for composition
, we adopted either a normal solar abundance @xcite , a wolf - rayet composition @xcite , or a pure gas of ionized hydrogen . for the spectrum , we adopted broken power laws with a low - energy cutoff , as described in the previous subsection .
the results for @xmath74 are displayed in figure 2 .
we find that the composition of the gas has little effect .
we find that the flux around one kev ( in the frame of the host galaxy ) is very important , due to the large cross sections associated with bound - free interactions .
unfortunately , the flux for energies below 7 kev are not known and we can only assume plausible spectral shapes , as discussed in the previous section . our favorite model of the final velocity as a function of the radial distance from the burster
is represented by the thick line in figure 2 . to accelerate gas from rest to 3000 km
s@xmath1 implies that the gas is at a distance of 0.2 parsec .
if the gas were already expanding at a velocity of 2500 km s@xmath1 , then the radial distance of the gas would be @xmath4 0.5 parsec .
gas initially at rest at @xmath4 0.5 pc , would be accelerated to @xmath4 500 km s@xmath1 , corresponding to one of the observed absorption components .
if the cloud has a substantial extent along the line of sight , then the inner edge will be accelerated to a higher velocity than the outer edge .
this would result in a broad line , instead of the narrow lines observed .
the radiative acceleration mechanism thus requires that the intervening cloud be physically narrow .
this could arise from a small cloud ( as a clumpy part of a wind ) or from a thin shell ( presumably created by the progenitor ) . for quantitative limits ,
the width of the carbon lines appears to be less than 400 km s@xmath1 , for which the dominant absorbing region in the shell must be thinner than 7% of its radius provided the gas is accelerated from rest to around 3000 km s@xmath1 by the burst radiation .
if the gas already had an initial velocity of 2000 km s@xmath1 , then the dominant absorbing region in the shell must be thinner than 21% of its radius .
if an absorbing gas has a wind - like density distribution ( that falls off as the inverse - square of the radial distance ) , then 50% of the absorption will come from between the inner edge of the absorbing region to twice that radius .
thus , for absorption from a smooth wind , the radiative acceleration must be small compared to the initial wind velocity to avoid excessively broad absorption lines .
however , if the absorbing gas has a significant clump superposed on a wind - like structure ( see 5 ) , then such a clump may be sufficiently thin to form a narrow absorption line despite a large radiative acceleration even in a wind .
this acceleration of gas close to the burster must occur , the only question is the size of the effect .
this will depend primarily on unknown details of the low energy photons in the burst spectrum after they have passed through material at small radial distances . for reasonable assumptions ( @xmath75 kev ) , the radiative acceleration will induce velocity changes of 1000 km s@xmath1 or greater for gas within a fraction of a parsec from the burster .
this accelerated gas must include that which is creating the afterglow , as well as the dense gas interior to the afterglow shock that represents the inner region of any wind that was driven off the progenitor .
wolf - rayet stars are very hot ( 30,000 - 70,000 k ) stars shedding a thick wind for which the characteristic spectra involve bright and broad emission lines with p cygni profiles .
they are highly evolved stars , the main sequence masses of which were larger than 35 @xmath76 or so . for the most massive stars ( @xmath77 ) the evolutionary path is from an o main sequence star to a luminous blue variable to a wolf - rayet star of the wn class to a wolf - rayet star of the wc class to a supernova collapse , while a less massive star ( @xmath78 ) progresses from an o main sequence star to a red supergiant to a wolf - rayet star of the wn class to a supernova collapse .
significant evidence and theoretical models support the hypothesis that long - duration @xmath0-ray bursts are caused by the collapse of massive stars @xcite , consistent with the notion that the immediate progenitors of bursts are wolf - rayet stars .
each stage in the evolution of massive stars has a distinct wind @xcite .
the o main sequence star wind is characterized by a velocity of @xmath79 km s@xmath1 and a mass loss rate of order @xmath80 to @xmath81 which lasts for a few million years .
the luminous blue variable phase has a massive ejection ( at a rate of up to @xmath82 ) over a short time ( roughly 10,000 years ) with moderate velocities ( around 300 km s@xmath1 ) .
the red supergiant wind is characterized by a velocity of 10 - 100 km s@xmath1 and a mass loss rate of up to @xmath83 which lasts for around 200,000 years .
the wolf - rayet wind is characterized by velocities of 1000 - 3000 km s@xmath1 and a mass loss from @xmath84 to @xmath85 over a lifetime of fractions of a million years .
the interactions between the winds in each successive stage of evolution will create shells with various properties @xcite .
the o main sequence star will blow a bubble into the surrounding ism that has a thin shell and a slow expansion velocity at a distance of order 30 pc .
the massive ejection of a luminous blue variable will produce a thick shell that expands with a velocity of around 300 km s@xmath1 .
the red supergiant wind will bunch up at its outer edge and expand with a velocity of 10 - 100 km s@xmath1 .
the wolf - rayet wind blows a bubble in the slower and denser wind around it , creating a thin shell that expands with a velocity of order 500 km s@xmath1 .
this bubble will overtake the slower shell in @xmath86 years and coalesce to form a combined turbulent shell . at the time of the core collapse of the star , the inner several parsecs will consist of the wolf - rayet wind expanding at 1000 - 3000 km s@xmath1 , surrounded by a dense and turbulent shell expanding at perhaps 500 km
s@xmath1at a distance of around 10 pc , surrounded by a relatively low density region evacuated by the o main sequence star wind , all inside a geometrically thin bubble of material piled up from the ism .
the winds of wolf - rayet stars are likely to be anisotropic , and the absorption lines in grb 021004 were created by only a small geometrical cross section , presumably along the rotation axis of the progenitor star .
can this wind structure around a very massive star account for the observed velocity components in the absorption lines of grb 021004 ?
the blue shifted 3000 km s@xmath1 component would arise from the dense inner wind driven during the wolf - rayet phase .
the o main sequence star wind also has a velocity of 3000 km s@xmath1 , but its density has been greatly reduced by its expansion into space so that it should produce only negligible absorption . the requisite density in this radius range
could also be provided in principle by the slower wind of a lower mass star with lower mass loss rate ; this case would subsequently require more radiative acceleration .
if the density in the wind falls off as the inverse - square of the radial distance from the star , the absorption from the wind will be dominated by its inner region , say 0.20.4 pc .
this means that the absorption lines would primarily sample a wolf - rayet wind emitted over a brief time interval somewhat before the core collapse . over its lifetime ,
a wolf - rayet wind is roughly constant in velocity , while the wind velocity across the 0.20.4 pc region should be even more precisely constant .
this could produce a narrow absorption line just as is observed .
the blue shifted 560 km s@xmath1 absorption component in grb 021004 could arise from the dense shell created at the outer edge of a wolf - rayet wind .
this shell is expected to be composed of geometrically thin structures that form a turbulent front width of which is roughly a tenth of its radius which expands at roughly 500 km s@xmath1@xcite .
the velocity structure within the shell is chaotic and will lead to some broadening of the resultant absorption lines .
thus , expected wolf - rayet wind structures can produce the observed 3000 km s@xmath1 and 560 km s@xmath1 components . what are the expected column densities in the wolf - rayet winds ?
for a mass loss rate of @xmath87 , a wind velocity of 3000 km s@xmath1 , and an average atomic weight equal to that of helium , a wolf - rayet wind will produce a column density of roughly @xmath88 .
the total column density of the shell at the outer edge of a wolf - rayet wind is uncertain , but it is likely to be roughly the mass ejected @xcite in the red supergiant phase ( @xmath4 20 @xmath76 ) or the luminous blue variable phase ( @xmath4 10 @xmath89 ) spread over a shell of radius 10 pc . for an average atomic mass like that of helium , the total column density will be of order @xmath90 atoms @xmath50 or greater .
the two shell components thus have comparable column densities , just as the two absorption lines have comparable equivalent widths .
are these column densities enough to produce significant absorption lines ?
for the oscillator strengths of and an assumed velocity dispersion of order 100 km s@xmath1 , a column density of @xmath91 for atoms in their ground state will produce an absorption line with an optical depth of unity . with the total column density for a component of a wolf - rayet wind at around @xmath92 atoms @xmath50 , the carbon line would produce a significant feature provided that in the ground state is more than @xmath48 of the nucleons in the wind
. a large fraction of the ejecta from wolf - rayet stars will be carbon .
the various ionization states will all spend almost all the time at their ground level because the decay times are of order @xmath93 seconds .
the question of calculating the fraction of carbon that is ionized to any particular degree
( by the radiation from the burst and from the progenitor ) is a difficult problem both for the physics and for the lack of knowing the conditions of the gas ( 3.2 ) .
nevertheless , it is plausible that some significant fraction of the carbon will be three - times ionized . for one - tenth of the carbon being
three - times ionized , the total column of atoms capable of producing the line in the wolf - rayet wind will be @xmath94 atoms @xmath50 .
for such a case , the line will be highly saturated .
this estimate has many uncertainties , but it is clear that wolf - rayet winds can have sufficient carbon to create the observed absorption lines .
wolf - rayet stars often show p - cygni line profiles , with the emission component arising from portions of the shell off the line of sight .
if the wind structure described above exists around grb 021004 , we do not expect to see p - cygni profiles .
there are two reasons for this .
the first is that the @xmath0-ray burst radiation is strongly beamed by the jet , so most of the shell off the line of sight is not illuminated .
the second reason is that a flash of illumination on the shell will produce emission lines that are delayed by an amount that varies with position in the shell . for typical shell sizes ,
the emission component would be spread out over many years and will become too faint to be observable . thus the lack of p - cygni profiles in the high - excitation lines of grb 021004 is not an argument against their origin in a wolf - rayet wind . what is the lower limit on the mass of the original star such that it can still produce the required dense wind
? o - type stars with masses of @xmath95 can produce dense winds with surrounding shell structures @xcite , and hence might account for the observed lines in grb 021004 .
these stars can have wind velocities up to 3000 km s@xmath1 but with mass loss rates that might be smaller than allowed by our analysis of the spectral energy distribution ( see 4 ) .
the evolution of such stars would be from an o main sequence star to a red supergiant to a blue supergiant with a core collapse at some point .
while such a progenitor can not be called a wolf - rayet star , it is still a massive star surrounded by shells and hence could be the progenitor of grb 021004 .
since the clumpy density profile is so uncertain , the only rigorous lower limit we can place on this process is that for core collapse itself , that is @xmath96 .
for instance , a star that blew up as a red giant might have a carbon - rich wind and hence have the necessary density to give the column depth of .
such a structure would have to rely on radiatively accelerated clumps of matter at @xmath4 0.5 pc to produce the lower velocity features in the absence of a shell generated by the blue / red wind boundary .
we now have two scenarios that can explain the velocity and ionization structure of the high - excitation absorption lines seen in grb 021004 .
the first scenario postulates two gas shells or clumps initially nearly at rest at distances of @xmath26 pc and @xmath97 pc that are radiatively accelerated to velocities of 3000 km s@xmath1and 560 km s@xmath1 , respectively , as well as ionized by the flash from the burst itself .
the second scenario proposes that the progenitor is a wolf - rayet - like star with a wind velocity of 3000 km s@xmath1 that is surrounded by a shell with velocity 560 km s@xmath1 , the gas of which is ionized by the flux from the progenitor .
these two scenarios are actually part of a continuum of similar scenarios where the ionizations and velocities vary from being caused entirely by the progenitor to being caused entirely by the burst . for example , in the second scenario , the original wind might have started at a lower ionization state at a velocity of 1500 km s@xmath1 and been accelerated and ionized to the final values by the burst . in all these related scenarios , the absorption lines with near zero velocity are caused by chance clouds along the line of sight as far as 2500 pc from the burster that have been ionized in part by the burst radiation . in the first scenario
, we have to ask why a @xmath0-ray burst progenitor would have two geometrically - thin gas clouds at distances of @xmath26 pc and @xmath98 pc ? this is improbable unless there is some causal connection between the clouds and the burster .
the presence of two geometrically - thin clouds along the line of sight could be caused by two shells around the progenitor or by some number of clumps of gas around the progenitor . in the second scenario
, we also have shells of gas centered on the progenitor .
we conclude that the progenitor had shells or clumps of gas centered on it and associated with it .
models where the @xmath0-ray burst arises from the collision of two compact objects in close orbit are substantially less likely because these have no ready supply of gas close to the system .
the presence of shells is an unusual situation for stars , as few have multiple shells , although clumps are more likely .
the only stellar objects with multiple shells that have any realistic hope of being associated with @xmath0-ray bursts are supernovae and massive stars .
any shell from a supernova would have to be pre - existing ( as in the supranova model of vietri & stella , 1998 ) since the afterglow jet would be far outside any normal supernova ejecta if the supernova and burst exploded simultaneously .
@xcite point out there are difficulties with producing narrow lines with the observed equivalent width from an old supernova remnant .
the narrowness of the absorption lines would require that the ejecta has already cooled and condensed into filaments .
the question is then the likelihood of getting multiple filaments of greatly different velocity on our line of sight and the column density of any such filaments .
ideally , the absorption lines from the @xmath2 system can tell us about the abundances of carbon , silicon , hydrogen , and nitrogen in the innermost shell ejected by the progenitor ; and this can then tell us about the progenitor itself . for example , if the composition is that of a wolf - rayet shell of a particular type ( perhaps with some swept up ism material ) , then we would even get a good idea of the mass of the progenitor .
unfortunately for this program , we have few lines from the @xmath2 system ( see the many lines for grb 010222 , @xcite ) . also , the likelihood that our unresolved lines are saturated implies that the measured equivalent widths will be difficult to convert to abundances .
nevertheless , a full analysis based on our data or the data of , for example , @xcite , might place useful limits on the abundances of the shell and its possible origin .
an example of such an analysis can be made with respect to the carbon - to - nitrogen abundance .
both atoms have similar ionization potentials and should have somewhat similar ionization fractions into either or , the oscillator strengths of the two resonance lines are similar , and the cosmic abundance of nitrogen is down from that of carbon by a factor of three .
so simplistically , we would expect the line to have an equivalent width that is roughly three times deeper than that of , whereas ( see table 1 ) we do not see any nitrogen lines and the limit on the equivalent widths is a factor 4.6 . with the likelihood that the line has a saturated core , the implication is that nitrogen is significantly under - abundant compared to carbon .
this would appear to rule out wn type wolf - rayet stars .
the lower mass wolf - rayet stars undergo core collapse while the wind is still nitrogen rich @xcite , whereas the higher mass wolf - rayet stars have core collapse after the wolf - rayet has turned from a wn to a wc type @xcite .
thus , the lack of a line in the spectrum of grb 021004 appears to imply that the burst progenitor must be amongst the more massive of wolf - rayet stars .
alternatively , the progenitor might be a star for which the original mass was too low ( @xmath99 @xmath100 ) to allow it to become a wn - type star , yet it would still have fast and dense winds that are not nitrogen - rich . the lack of molecular hydrogen absorption can be used to place further limits on the density of the gas near the burster . in the model of @xcite , the column density of excited
@xmath101 , @xmath18 , arises from an ionization / dissociation front the dynamics of which imply that there are natural values of the column depth . for example , an increase in the burst fluence would produce a similar front ( only deeper into the molecular cloud ) with a similar @xmath18 , while an increase in the molecular density would produce a front with a narrower width yet with a similar @xmath18 .
the only way to produce a smaller column density is to have uv fluxes that are not sufficient to establish a proper ionization / dissociation front in the molecular hydrogen . this could arise if the molecular cloud were at a great distance ( say , 1 kpc ) from the burster or if the burster were near the edge of the molecular cloud .
draine calculates that a typical burst will destroy molecular hydrogen in gas with a hydrogen column density of @xmath102 ( corresponding to @xmath103 or so ) . for densities from 10010000 @xmath104
, the burst will destroy the molecular hydrogen that could serve as an absorber if it is within roughly 2200 pc of the edge of the cloud .
for the general case of the burst erupting well inside a molecular cloud , the @xmath18 will be the value used to calculate the template ( see section 2 ) , and this is rejected by our spectrum .
we conclude that the lack of molecular hydrogen absorption lines in our spectra indicate that the @xmath0-ray burst is not inside a molecular hydrogen cloud unless it is within @xmath42 - 200 pc from its edge .
we have constructed broad - band fluxes from the radio to the x - ray for several epochs ( see figure 3 ) with the data collected from the gcn notices .
the major features of the radio emission are : @xmath105 at 0.53.5 days the 15 ghz light - curve rises as @xmath106 , @xmath107 at 1.5 days the 1586 ghz spectrum is @xmath108 , appearing flat from 86 to 230 ghz , and @xmath109 at 5.7 days the 1.48.5 ghz spectrum is @xmath110 .
the last two features indicate that the synchrotron self - absorption frequency @xmath111 is not significantly above 15 ghz at 1.5 days and not much below 8 ghz at 5.7 days , which would suggest a wind - like medium ( for which @xmath112 ) instead of a homogeneous environment ( for which @xmath111 is constant ) .
however , with only a few measurements and the possibility of significant interstellar scintillation affecting the lower frequency ( under 15 ghz ) observations , it is premature to favor a wind - like medium based on the apparent decrease of @xmath111 over a factor of 4 in time .
the rise of the 15 ghz emission is consistent with both the expectations from a homogeneous medium ( where @xmath113 irrespective of the location of @xmath111 ) and a behavior intermediate to the asymptotic @xmath114 and @xmath115 expected for a wind - like medium . within the framework of relativistic fireballs , consistency between the decay rate of the optical emission
( @xmath116 with @xmath117 ) and spectral slope ( @xmath118 with @xmath119 ) requires that the cooling frequency @xmath120 is below or within the optical range , as in this case the expected relationship ( @xmath121 , irrespective of the type of external medium ) is satisfied .
furthermore , that @xmath120 is below optical is also consistent with the equality of the optical and x - ray decay indices ( @xmath122 ; sako & harrison 2002 ) and spectral slopes ( @xmath123 ) .
we note , however that the difference between the observed optical and x - ray decay indices , @xmath124 , is also consistent ( even if only marginally ) with the theoretically expected value @xmath125 in the case where @xmath120 is above the optical domain but below x - rays , and that the intrinsic optical spectral slope @xmath126 can be smaller than observed if there is a significant dust reddening within the host galaxy .
therefore , the cooling frequency could be above the optical domain , in which case one expects that @xmath127 for a homogeneous medium , and @xmath128 for a wind - like medium .
furthermore it is also expected that @xmath129 , which favors a homogeneous medium when compared with the above @xmath126 inferred from the optical decay .
concluding , the currently available radio and optical data for grb 021004 do not constrain sufficiently the type of external medium , homogeneous or wind - like .
however , the average density of the medium or the mass - loss rate to wind speed ratio can be determined if the cooling frequency lies within the optical range at about 2 days , as suggested by matheson et al 2002 . together with the optical measurements , the 15 , 86 and 230 ghz fluxes reported by pooley ( 2002 ) and
bremer & castro - tirado ( 2002 ) indicate that , at 1.5 days , the peak of the spectrum of grb 021004 is @xmath130 mjy , at a frequency between @xmath131 and @xmath132 hz . adding that @xmath133 ghz at a few days , these four constraints on the synchrotron spectrum are sufficient to determine the external medium density , fireball isotropic - equivalent energy , and fractional energies in electrons and magnetic fields .
we obtain an external density in the range @xmath134 for a homogeneous medium , consistent with the values found by panaitescu & kumar ( 2002 ) for other grb afterglows . for a wind - like medium ,
we find that density of the wind corresponds to that resulting from a mass - loss rate to wind speed ratio in the range 1 to several times @xmath135 km s@xmath1 ) , within the range expected for wr stars .
we have constructed light curves and color curves for grb 021004 from data published in the gcn notices ( see figure 4 ) .
overall , the light curve behaves approximately as a @xmath136 power law which is normal for many afterglows
. however , there are many significant bumps and wiggles about any simple single power law .
to help visualize these bumps , the middle panel of figure 4 shows the r - band light curve after a @xmath136 power law has been subtracted , and we see three bumps with peaks at 0.14 , 1.1 , and 4.0 days after the burst .
the v - r color is plotted in the lower panel of figure 4 .
the color appears to remain essentially constant from 0.2 to 10 days .
afterglow models suggest that light curves should consist of broken power laws , so there is an urge to try to interpret figure 4 in terms of broken power laws .
perhaps the basic behavior is @xmath136 with bumps and some separate mechanism for the dimness at the earliest times .
alternatively , perhaps the basic behavior is a @xmath137 power law up until a break at 3 days to a @xmath138 power law with a particularly strong bump from 0.10.7 days . for a third alternative ,
perhaps the basic behavior is a @xmath138 decline over the entire time range except that some mechanism inserts more energy three times during the decline ( at 0.14 , 1.1 , and 4.0 days ) . in general , @xmath0-ray burst afterglows have simple power law declines in brightness , often with a break ascribed to collimation of the outflow .
nevertheless , many @xmath0-ray bursts have bumps in their afterglows .
for example , the optical brightness of the afterglow of grb 970508 increased by 1.5 magnitudes 12 days after the @xmath0-ray burst @xcite .
grb 970508 was seen by the bepposax satellite to have a significant brightening in its x - ray light curve starting at 16 hours after the burst @xcite .
grb 000301c displayed an achromatic bump that peaked around 4 days after the burst that has been ascribed to gravitational microlensing by @xcite .
in addition , three @xmath0-ray bursts with bumps around 30 days after the burst have been ascribed to an underlying supernova ( reichart 1999 , bloom et al . 1999 ; 2002 ; but see dermer 2002 ) or dust echos ( reichart 2001 ) . the presence of bumps in otherwise - smooth afterglow light curves has previously been attributed to microlensing , dust echos and supernovae .
now we have a burst with three bumps , and they can not all be due to microlensing , dust echos , or supernovae , so we have a proof that afterglows can makes bumps on various time scales for other reasons .
thus , the mere presence of a single bump can only provide poor support for any particular proposed cause unless other possibilities can be excluded by some means .
for example , the red color of a bump ( compared to the underlying afterglow ) can not be taken as substantial evidence of a supernova origin until the cause and color - distribution of other classes of bumps are known .
there are two generic scenarios ( in addition to microlensing , dust echos , and supernovae ) for producing variability in the afterglow emission : an inhomogeneous circumburst medium and non - uniform properties of the @xmath0-ray burstejecta ( see also nakar , piran & granot 2003 ) .
inhomogeneities in the external gas could be either in the form of clumps of denser material @xcite , which enhance the dissipation of the kinetic energy of the @xmath0-ray burst remnant and , implicitly , the afterglow brightness , or in the form of a stratified medium , consisting of shells of various densities @xcite , perhaps produced by fluctuations in the wind blown by the @xmath0-ray burst progenitor not long before the release of the @xmath0-ray burst ejecta .
non - uniformities in the @xmath0-ray burst ejecta could be manifest either as a range of initial lorentz factors within an impulsive ejection or as a non - isotropic distribution of the ejecta energy per solid angle . in the former case ,
the slower part of the outflow catches up with the faster part as the latter is decelerated , enhancing the afterglow kinetic energy and , hence , its brightness . in the latter case ,
the regions of higher angular energy density are bright spots on the ejecta surface , moving slightly off the direction toward the observer , so that fluctuations in the afterglow emission are produced as the spots decelerate and the widening cone of their relativistically beamed emission starts to include the direction toward the observer .
the variable - lorentz - factor and anisotropic outflow scenarios have been used to explain the brightening of the afterglow of grb 970508 by @xcite and @xcite , repectively .
an energy input that is extended in time prior to the brightening at 0.14 days could explain the shallower decline prior to 0.14 days , while an inhomogeneous shell with a density decreasing with radius @xcite could account for the steeper light - curve decay after 0.1 days in the stratified medium scenario .
the faster decline after a brightening is consistent with the expectations from scenarios involving dense clumps or bright spots , where the contribution of the clump or spot to the total afterglow emission decreases in time as the fireball decelerates , and more of the uniform ejecta surface becomes visible for the observer . in the latter two scenarios where the afterglow emission from clumps or spots declines with time
, the afterglow light curve should eventually recover its initial decay rate . in this context , the observed light curve is consistent with the proposal that the @xmath137 fall - off seen in the first 15 minutes reappeared at 0.7 days , just about the time the spectra reported here were obtained , and lasted until 13 days after the burst , when the light curve steepened to @xmath139 .
if this interpretation is correct , grb 021004 would display the shallowest light - curve break ever observed .
the change in the break index would be @xmath140 , suggesting that the break is caused by the passage of the cooling break through the optical domain @xcite .
the existence of a break in the optical spectrum at 13 days is supported by the spectroscopic observations by @xcite and this paper ( see 2 ) , that the blue part of the optical spectrum reddens between @xmath141 and @xmath142 days .
this interpretation is consistent with a break frequency that decreases in time , implying that the external medium is homogeneous rather than wind - like . in the case of a wind ,
the cooling frequency should increase at least as fast as @xmath143 @xcite .
we note , however , that the break in the optical spectrum seems too sharp in wavelength to be due to the cooling break ( across which the slope of the power - law spectrum changes by 0.5 ) , and that the spectral changes may not sufficiently affect the r - band light - curve at a few days to explain the r - band light curve behavior .
the radial distance of the jet from the center of the progenitor as a function of the observer s time should be correlated with the location of clumps and the epoch of the light curve bumps .
this can be modelled given both the broad - band spectrum and the light curve by the methods presented in @xcite .
unfortunately , only preliminary input numbers are available ( for example , no jet break has yet been definitely timed ) and details of the external medium are not known ( for example , whether or not the average density falls off like a wind ) .
nevertheless , it is possible to give estimates that are useful for producing a reasonable picture . for this
, we adopt the isotropic equivalent energy of the fireball equal to @xmath144 erg @xcite .
if the jet is expanding into an external medium with a wind - like structure , then we found in the previous section that the mass - loss - rate to wind - speed ratio is roughly @xmath145 per 1000 km s@xmath1 . in this case , the radius of the jet will be @xmath146 pc , with @xmath41 measured in days .
if the jet is expanding into a homogenous medium of constant density , then we found in the previous section that the density is roughly 30 @xmath147 . in this case , the radius of the jet will be @xmath148 pc .
if the light curve bumps are due to density enhancements in the surrounding medium , then we can estimate the approximate radial distances from the burst and over - densities for each clump . for peaks in the light curve at 0.14 , 1.1 , and 4.0 days , we find approximate radii for the density clumps of 0.022 , 0.063 , and 0.12 pc , respectively , for a wind - like medium .
we get similar results for expansion into a homogenous medium with parameters as derived above .
the excess light in the light curve bumps will scale roughly as the square root of the over - density compared to the background " density , for the case where the clump is relatively large compared to the visible region ( as for a shell ) . from figure 4 , the extra light in each of the bumps is roughly 0.6 , 0.3 , and perhaps 0.3 mag ; this corresponds to over - densities of roughly 50% , 10% , and 10% .
if the bumps are due to clumps , the broad - band spectral slopes should remain the same and thus the colors would not change across the bump , as observed . at some time
the afterglow jet will collide with the gas that is causing the @xmath2 absorption .
when this happens , we would expect that the afterglow will brighten significantly if the gas is in a relatively narrow shell .
we also expect that the associated absorption lines will disappear .
we note that our late het spectrum still showed the @xmath149 line with moderately low significance at a time of 4.84 days after the burst when the jet would have reached @xmath150 pc . in principle
, we can predict the date at which this bump in the afterglow decline should occur . in practice
, such a prediction will have substantial uncertainties , primarily in the exact radius of the shell and the density structure of the medium inside the shell .
in the case where a shell of material has been radiatively accelerated to 3000 km s@xmath1 , the shell will have a radius not much larger than 0.1 pc and it should be over - run sometime between roughly 240 days after the burst . in the case where the absorption is due to a wolf - rayet wind that extends out to several parsecs , the 3000 km s@xmath1 component will be completely eliminated only after many years .
the absorption will slowly diminish during the first few months after the burst as the jet catches up with the densest inner part of the wind . in summary ,
the absorbing gas might be eliminated during the early days after the burst ( an event that could be directly observable ) or it might take many years to be eliminated ( and be totally unobservable ) .
the grb 021004 spectrum displays high - ionization and high - velocity absorption lines .
we argue that these lines are likely to have arisen from shells of gas around the progenitor star , and further , that these shells plausibly come from a progenitor that was descended from a massive star . the gas in these shells was partially ionized and accelerated by the photons from the burst .
we have used the optical spectrum , the broad - band spectrum , and the light curve to identify six components in the medium external to the @xmath0-ray burst .
the @xmath2 absorption lines arise from a shell at a radial distance of less than a parsec that is moving at a velocity of 3000 km s@xmath1after being ionized and accelerated by the burst radiation .
we identify this component as being the wind from the progenitor star , perhaps a wolf - rayet star with a mass loss rate of @xmath151 or a lower mass star wind accelerated by the bust .
this same component provides the external medium with which the jet is colliding to produce the afterglow light .
this inner component is probably significantly clumped in order to prevent complete ionization and to facilitate recombination , thus allowing a significant column depth of .
the second component is associated with the absorption lines moving 560 km s@xmath1 with respect to the host galaxy velocity .
we identify this component as arising from a shell of material plowed up by the wolf - rayet - like wind at a radius of order 10 parsecs .
alternatively , this feature could be a shell or clump at @xmath4 0.5 pc accelerated by the burst .
the third component gives rise to the low - velocity high - excitation absorption lines which must come from a distance between tens and hundreds of parsecs from the burster .
this component might be from the shell created by the o main sequence star wind or it might be from chance ism clouds along the line of sight , perhaps part of the same star forming region that gave birth to the @xmath0-ray burstprogenitor .
the fourth , fifth , and sixth components are three density enhancements that occur in the wolf - rayet - like wind at radial distances of 0.022 , 0.063 , and 0.12 pc so as to produce the three bumps in the light curve .
these clumps could be related to the clumping we deduce for component 1 , that is , all could be associated with knots in the progenitor wind .
the results summarized in the previous paragraph offer a plausible description of the density structure near this burst .
what we see is a variety of density enhancements along the line of sight , with the evidence coming from bumps in the light curve decline and absorption lines in the spectrum .
this picture is apparently not universal for @xmath0-ray bursts , since only two out of nine bursts with well - observed afterglow spectra show both high - excitation and high - velocity lines .
the other bursts must have some mechanism for suppressing similar absorption lines .
a full analysis is needed for the @xmath12 lines of grb 021004 , with the possibility of determining the abundance of the progenitor s shell and hence to obtain information about the progenitor itself . as a start to this program
, we have shown that the nitrogen in the shell is under - abundant with respect to the carbon , and this demonstrates that the progenitor was not a wn - type wolf - rayet star and further that the progenitor was either one of the more massive wolf - rayet stars or a massive star that is below that required to make a wn - type star .
a further program for our community is to obtain high dispersion spectroscopy so as to be able to measure fine velocity structures and faint lines in the afterglow spectra of many bursts .
the most important result from our data and analysis is that a plausible way to create the high - excitation and high - velocity absorption lines is to have shells of gas surrounding the progenitor , and that a realistic way to achieve this is for the progenitor to be a star descended from a massive o - type main sequence star .
this strongly suggested connection to a massive star is much more direct than the moderate statistical association of @xmath0-ray bursts with parts of galaxies undergoing star formation and is much more convincing than attributing poorly observed bumps in afterglow light curves to some associated supernova event .
as such , we believe that the absorption lines in the grb 021004 spectra provide the first strong and direct argument for associating a normal @xmath0-ray burst with a massive progenitor star surrounded by shells .
we thank bruce draine , bev wills , greg shields , and d. j. hillier for their discussion on a wide range of topics .
this work was supported in part by nasa grant nag59302 and nsf grant ast-0098644 .
the marcario low resolution spectrograph is a joint project of the hobby - eberly telescope partnership and the instituto de astronomia de la universidad nacional autonoma de mexico .
the hobby - eberly telescope is operated by mcdonald observatory on behalf of the university of texas at austin , the pennsylvania state university , stanford university , ludwig - maximilians - universitt mnchen , and georg - august - universitt gttingen .
ccccc @xmath152 & si iv & 1393.8 & @xmath153 & 2.291 + @xmath154 & si iv & 1393.8 & @xmath155 & 2.321 + @xmath154 & si iv & 1402.8 & @xmath155 & 2.299 + @xmath156 & si iv & 1402.8 & @xmath157 & 2.326 + @xmath158 & c iv & 1548.2 & 1550.8 & @xmath159 & 2.296 + @xmath160 & c iv & 1548.2 & 1550.8 & @xmath161 & 2.325 + @xmath162 & al ii & 1670.8 & @xmath163 & 2.328 + @xmath164 & n v & 1238.8 & 1242.8 & @xmath165 & @xmath166 | we present spectra of the optical transient of grb 021004 obtained with the hobby - eberly telescope starting 15.48 , 20.31 hours , and 4.84 days after the @xmath0-ray burst and a spectrum obtained with the h. j. smith 2.7 m telescope starting 14.31 hours after the @xmath0-ray burst .
grb 021004 is the first burst afterglow for which the spectrum is dominated by absorption lines from high ionization species with multiple velocity components separated by up to 3000 km s@xmath1 .
we argue that these absorption lines are likely to come from shells around a massive progenitor star .
the high velocities and high ionizations arise from a combination of acceleration and flash - ionization by the burst photons and the wind velocity and steady ionization by the progenitor .
we also analyze the broad - band spectrum and the light curve so as to distinguish the structure of gas within 0.3 pc of the burster .
we delineate six components in the medium surrounding the @xmath0-ray burst along the line of sight : ( 1 ) the @xmath2 absorption lines arise from the innermost region closest to the burst , where the ionization will be highest and the 3000 km s@xmath1 velocity comes from the intrinsic velocity of a massive star wind boosted by acceleration from the burst flux . for a mass loss rate of @xmath3
, this component also provides the external medium with which the jet collides over radial distances 0.0040.3 pc to create the afterglow light .
( 2 ) a second cloud or shell produces absorption lines with a relative velocity of 560 km s@xmath1 .
this component could be associated with the shell created by the fast massive star wind blowing a bubble in the preceding slow wind at a radial distance of order 10 pc or by a clump at @xmath4 0.5 pc accelerated by the burst .
( 3 ) more distant clouds within the host galaxy that lie between 30 - 2500 pc and have been ionized by the burst will create the @xmath5 absorption lines .
( 4 - 6 ) if the three bumps in the afterglow light curve at 0.14 , 1.1 , and 4.0 days are caused by clumps or shells in the massive star wind along the line of sight , then the radii and over - densities of these are 0.022 , 0.063 , and 0.12 parsecs and 50% , 10% , and 10% respectively .
the immediate progenitor of the @xmath0-ray burstcould either be a wc - type wolf - rayet star with a high velocity wind or a highly evolved massive star the original mass of which was too small for it to become a wn - type wolf - rayet star . in summary ,
the highly ionized lines with high relative velocities most likely come from shells or clumps of material close to the progenitor and these shells were plausibly produced by a massive star soon before its collapse . |
stellar disks are known to exist around massive black holes ( mbhs ) in galactic nuclei , and are typically composed of relatively young stellar populations .
observations of the central parsec of the milky - way galaxy show more than 100 young wr , o and b stars @xcite , likely formed @xmath0 @xmath1 ago .
these stars are found between @xmath2 pc from the mbh , with @xmath3 of which reside in a disk - like structure , and are observed to have mean high eccentricity of @xmath4 .
stellar disks might therefore be a frequent phenomena in galactic nuclei .
such disks are thought to form following the infall of gaseous material , and its formation of a gaseous disk .
such a disk can later fragment to form a stellar disk ( * ? ? ? * and references therein ) . here
we explore the dynamical evolution of such stellar disks , and focus on disks similar to that observed in the galactic center .
@xcite have been the first to describe the behavior of a stellar disk around a mbh .
they used an analytic approach , and then verified and calibrated them through the use of n - body simulations .
the analytic model they devised includes the effects of two - body relaxation , and included simple multi - mass models ( two or three populations ) .
their n - body simulations also included a simplified stellar evolution .
later studies by @xcite , @xcite and @xcite performed additional n - body simulations and considered additional components and/or more realistic aspects .
@xcite included stellar binaries in the disk , the role of a putative intermediate mass black holes , and initially eccentric disks .
@xcite discussed the evolution of the stellar disk , mainly focusing on the effects of resonant relaxation @xcite , not considered here .
the influence of the stellar cusp was studied numerically by @xcite and @xcite , who performed n - body simulation for different models ( including models with cusp and models with two stellar disks ) ; they concluded that the cusp could provide an additional heating of the stellar disk , leading to faster eccentricity / inclination growth .
here we follow - up on the initial analytic work by @xcite , and extend it to include the role of a realistic detailed mass - function , the effects of stellar evolution , the impact of binary - heating and additional heating by a stellar cusp .
this approach allows us to better understand and identify the role played by each of these processes and components , as well as to study the long - term evolution of realistic disks , which is more difficult ( and computationally expensive ) to study using full n - body simulations .
we begin by describing the dynamical processes that contribute to the stellar disk evolution , including stellar scattering , mass segregation and binary - heating ( section 2 ) .
we then briefly review the structural components of the nuclear stellar cluster in which the disk is embedded , and consider various models for the mass function of the stellar disk , and the possible core - like or cuspy structure of the nuclear stellar cluster ( nsc ) .
the physical processes we consider as well as the nsc structural components are used to devise a multitude of possible models for the stellar disk evolution under diverse conditions , including considerations of stellar evolution ( section [ sec : models ] ) . we present the results of the short and long term evolution of the disks considered in these models in section [ sec : results : ] , and then discuss the results and their implications ( section [ sec : discussion : ] ) .
we also briefly mention other possible physical processes not included in our current modeling .
finally we conclude and summarize our results in section [ sec : summary ] .
in the following we briefly review several dynamical processes that contribute to the disk evolution .
we first follow @xcite in describing the two - body relaxation and mass - segregation ( dynamical friction ) of multiple stellar populations ( extended to an arbitrary number of stellar populations ) , and then consider the analytic description of the effects of binary heating on stellar disks , not discussed before in this context .
let us follow @xcite and consider a super massive black hole ( mbh ) with mass @xmath5 and @xmath6 single stars with mass @xmath7 orbiting the mbh in a initially thin stellar disk , with an external radius @xmath8 and an internal radius @xmath9 .
when @xmath10 the relaxation time of such a system is given by @xcite .
@xmath11 with @xmath12 the stellar density and @xmath13 ( depending only on the system geometry ; e.g. for a spherical case @xmath14 @xcite .
the disk density is given by @xmath15 where @xmath16 is the scale height of the disk .
we then obtain @xmath17 using this eq . and defining @xmath18 we can describe the evolution of the velocity dispersion of disk stars : @xmath19 with @xmath20 . in a disk of stars orbiting an mbh , the root mean square ( rms ) eccentricity of the stars
can be related to their velocity dispersion by ( * ? ? ?
* and references therein ) @xmath21 where @xmath22 is the keplerian orbital speed .
the rms inclination of stars is expected to be of the order of half the rms eccentricity value . for comparison with observations of the eccentricities of stars in the galactic center we therefore present our results of the rms eccentricity ( and inclination in some cases ) rather than the velocity dispersion .
[ fig : sigma - evol - simple ] shows the evolution of the rms eccentricity of stars in the stellar disk , using a similar setup as used by @xcite evolution of the velocity dispersion of stars in a stellar disk composed of single mass stars orbiting a mbh ; similar to @xcite . ] stellar scattering had also been considered for multi - mass stellar populations in the dispersion dominated regime @xcite
. in this regime , for two mass populations , suggest that the treatment of the problem only depends on the amplitude of the velocity dispersion ( @xmath23 ) of the lighter objects , with mass @xmath24 , compared with the hill velocity of the more massive components , with mass @xmath25 .
the hill velocity for the more massive stars is defined as @xmath26 where @xmath27 .
the dispersion dominated regime occurs when @xmath28 . in this regime
the speed of the single stars is similar to the velocity dispersion and scattering encounters are well approximated by two body dynamics . in the shear dominated regime ,
not discussed here , it is necessary to take into account the tidal gravity of the central black hole . when the velocity dispersion of the low - mass stars satisfies @xmath29 , with @xmath30 the escape velocity from massive stars , the exchange of momentum between the heavier and lighter bodies occurs through collisionless gravitational deflections , where gravitational focusing is important .
considering only two stellar populations composed of @xmath31 and @xmath32 stars with stellar masses @xmath25 and @xmath24 , and velocity dispersions @xmath33 and @xmath23 , respectively , we obtain @xcite : @xmath34 @xmath35 where @xmath36 and @xmath37 for @xmath38 and 2 , and @xmath39 are the appropriate coulomb logarithms for the two stellar populations , which depend on the respective scale height of the disk .
@xmath40 is related to the scale height at which interactions between the two stellar species occur ( see * ? ? ? * for further details )
the constant @xmath41 is related to the constant @xmath42 , and also depends on the geometry of the system .
i.e for a three dimensional system @xmath43 ( for a detailed analysis please see at @xcite ) ; here we adopt the same value .
we can extend such a model to an arbitrary number of different stellar species , @xmath6 .
each species adds an additional coupled equation and contributes a coupling term to each of the other equations . for the general case
the time evolution of the velocity dispersions for each of the stellar species is given by a set of coupled equations @xmath44 for @xmath45 and @xmath46 ( with @xmath47 ) . from analyzing these set of equations we note that for two species , @xmath48 and @xmath49 for which @xmath50 , the population of @xmath51species particles dynamically heat the population of the @xmath49-species ones ( leading to increased velocity dispersion ) and vice versa ( a less energetic population dynamically cools a more energetic one , dumping its velocity dispersion ) .
the two - body relaxation of disk stars due to interaction with the background stars of the nuclear cluster can be modeled in a similar method to the two - body relaxation by different stellar populations discussed above . in this case , the cluster stellar population is considered as a distinct population with different properties .
the disk stars interact with those cluster stars which go through the disk , i.e if the volume of the disk is @xmath52 , the number of clusters stars considered is the nuclear cluster stellar density times this volume .
the velocity dispersion of this isotropic cluster population is @xmath53 @xcite with @xmath54 the keplerian velocity around the mbh .
we now use the same type of two - population interaction term in eq .
[ eq : dsigmaj_dt ] to account for the effects of the cluster stars on the dis stellar population .
note that the cluster stellar population is much greater than the disk population , and can be assumed as a constant `` heat '' bath .
any stellar population is likely to contain some fraction of binary stars .
an encounter between single stars and binaries could lead to an energy exchange between the binary orbital energy and the kinetic energy of the stars , i.e. , energy can be exchanged between the inner degree of freedom of the binary and the outer external kinetic energy of the stars ; thereby binaries can assist in dynamically heating the stellar disk .
let us consider a binary system composed of stars with masses @xmath55 and @xmath56 , center of mass velocity @xmath57 , and orbital velocity @xmath58 with respect to the center of mass of the system ( we assume a circular binary for simplicity ) .
the total energy of the binary can be expressed as @xcite : @xmath59 where @xmath60 is the reduced mass .
the sum of the first two left element define the binding energy , @xmath61 , in terms of the semi - major axis @xmath62 binary - single encounters had been explored extensively , with the pioneering studies by @xcite and @xcite .
binaries can be divided into two dynamical categories ; `` hard binaries '' and `` soft binaries '' : when the kinetic energy of the intruding mass , @xmath63 , is such that @xmath64 , ( where @xmath65 is the binary total mass)a binary is considered to be hard ( where the lower the ratio the harder the binary ) ; it is considered to be a soft binary when @xmath66 .
@xcite and @xcite found that following a binary - single encounter soft binaries become softer and hard binaries become harder ( on average ) , a behavior typical termed as `` heggie s law '' .
the long term typical evolution of a binary randomly encountering stars in some dense environment can then be modeled using a simple approach . considering a binary in a field of incoming stars with typical mass @xmath67 and particle density @xmath68 , the averaged evolution of the binding energy of the binary due to many encounters
is given by @xmath69 where @xmath58 the typical relative velocity of the intruders , and @xmath70 the cross section of the single - binary collisions .
we may now consider the contribution of a stellar population @xmath71 with stellar density @xmath72 which is composed of stars with typical mass @xmath73 , to the binary evolution ; when the velocity distribution of incoming stars is maxwellian and taking eq .
[ eq : de_b_dt ] can be expressed as @xcite : @xmath74 where @xmath65 , @xmath75 .
@xcite made use of n - body simulations to study the dependence of @xmath76 on the intruder mass ( @xmath73 ) and the binary mass ( @xmath77 ) in the mass - ratio range @xmath78 $ ] and found that the the empirical functional dependence can be formulated by @xmath79 with @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 and @xmath87 . for a single star
the average kinetic energy is @xmath88 with @xmath89 the velocity dispersion .
taking @xmath90 ( encounter velocity @xmath91dispersion velocity ) and applying energy conservation for a system of binary stars with density @xmath92we get @xmath93 eqs .
[ eq : de_v_dt_f ] and [ eq : equal_energy ] give @xmath94 with @xmath95 we can now use eq .
[ eq : binary_heting ] as the binary - heating term to be added to the dynamical evolution equation for the stellar disk populations ( eq . [ eq : dsigma_dt ] ) ; allowing us to model the binary - heating contribution .
we are now in a position to asses whether binary heating can play an important role in heating the stellar disk .. for the simplest case of a disk with only one population of same - mass stars and a binary fraction @xmath92 , the velocity dispersion evolution equation is obtained by summing the scattering term and the binary heating to get ( taking @xmath96 ) @xmath97 or after some math @xmath98,\label{eq : bin - vs - single}\ ] ] where @xmath99 is the number of binary stars , and @xmath100 we can now compare the relative contribution by single - star scattering and binary heating through the comparison of the two terms in eq .
[ eq : bin - vs - single ] .
if we consider some typical initial conditions in the galactic center , taking @xmath101 ( as the disk heats - up the scale height grows but with only a small corresponding change in @xmath102 due to the logarithmic dependence ) , @xmath103 and a binary fraction of 2/3 , @xmath104 , we find that the relative contribution of binary heating is only one tenth of the contribution due to single star scattering . this can be seen explicitly in the evolution of the velocity dispersion shown in fig .
[ fig : binaries-2 ] .
the left panel shows how the exclusion of the binary heating contribution ( for a disk in which @xmath105 ) makes only a very small difference ( see also the effect of accounting only for binary heating ) .
we can therefore conclude that although the binaries binding energy could be larger that the kinetic energy of the disk , the rate of extraction of this energy and its exchange into the disk kinetic energy is negligible , at least for the relatively thin disks considered here .
we can combine the stellar disk contributions described above to obtain the full set of equations describing our model , including multi - mass populations , binaries and nuclear cluster stellar population .
this is summarized below ; @xmath106 where 1 .
@xmath107 2 .
@xmath108 , 3 .
@xmath109 4 .
@xmath110 5 .
@xmath111 , where we consider the effective region to affect the stars to be between @xmath112 and@xmath113 and 6 .
@xmath114 7 .
@xmath115 , where @xmath116 and @xmath117 is semi - major axis of the binary . 8 .
@xmath118 and table [ tab : models ] for more information about the models ) .
[ fig : binaries-2 ] . in the top panel
is plotted the evolution of rms eccentricity through time , the dashed line is the model only considering auto - relaxation and the solid one is when binary heating is included . in the bottom panel
are ploted the the evolution of the rms eccentricity when only auto - relaxation ( dashed line ) or binary heating is considered ( solid line).,title="fig : " ] and table [ tab : models ] for more information about the models ) .
[ fig : binaries-2 ] . in the top panel
is plotted the evolution of rms eccentricity through time , the dashed line is the model only considering auto - relaxation and the solid one is when binary heating is included . in the bottom panel
are ploted the the evolution of the rms eccentricity when only auto - relaxation ( dashed line ) or binary heating is considered ( solid line).,title="fig : " ]
in the following we describe the observed properties of the cluster and the embedded stellar disk , and the models we use in our analysis . beginning with the analysis of @xcite
, observations have shown the existence of a young stellar disk near the mbh in the galactic center .
@xcite showed that many of the observed young o / wr / b - stars reside in a disk - like structure between @xmath119 pc from the mbh .
early measurements suggested the disk - stars revolve around the mbh in somewhat eccentric orbits.@xcite found an rms eccentricity of @xmath120 for the disk stars , but the most recent studies suggest a smaller value of @xmath121 @xcite .
the disk - stars appear to be young , forming only @xmath122 myrs ago @xcite .
the radial surface density profile of the disk can be described by a power - law distribution , @xmath123 , with @xmath124 in the range @xmath125 $ ] @xcite .
initially it was thought that @xmath126 of the young stars belong in the disk structure ; however , more recent studies suggest that only @xmath3 of these reside in the disk , while most of the stars have more spherical distribution , outside the disk @xcite . here
we consider only the evolution of disk stars . in most galactic environments
the initial mass function ( imf ) describes the distribution of stellar masses at birth .
analysis of observations ( @xcite , @xcite , @xcite ) shows an almost universal power law behavior , with @xmath127 , with a typical exponent of @xmath128 , for stars above one solar mass and up to @xmath129 @xmath130 .
the analysis done by @xcite based on spectroscopic survey made with sinfoni conclude that the present day mass function and the imf for the disk of young stars in the galactic center obey a power law , but the slope is flatter than the `` regular '' imf . in particular , they conclude that @xmath131 .
in such a `` top heavy '' imf , massive stars are much more frequent compared with the salpeter imf .
in our models we consider an initially thin cool disk ( @xmath132 and low initial eccentricities for the star , i.e. we assume an initially low velocity dispersion for the disk stars ) rotating at angular speed @xmath133 ( with @xmath54 the keplerian velocity ) .
we only consider the evolution of the disk stars and do not discuss the origin and evolution of the more spherical component of the young stars in the galactic center .
we studied three different types of disks : a single - mass disk , an initial mass function ( imf ) disk and disk with binaries .
we consider various mass functions for the stars in the stellar disk and the cusp .
these different models are summarized in the following .
[ [ single - mass - model ] ] single - mass model : + + + + + + + + + + + + + + + + + + + in the simplest , unrealistic model studied here , we consider a stellar disk composed of single - mass stars , similar to the basic models considered by @xcite , as a test - case for comparison .
[ [ top - heavy - initial - mass - function - model ] ] top - heavy initial mass - function model : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this model we assume the initial mass function of the disk stars is a top - heavy mass function with masses in the range @xmath134 @xmath135 , as suggested by ( * ? ? ?
* see discussion above ) .
[ [ salpeter - mass - function ] ] salpeter mass function : + + + + + + + + + + + + + + + + + + + + + + + in this case we assume the initial mass function of the stars in the stellar disk is a salpeter mass function with masses in the range @xmath134 @xmath135
. our knowledge about binaries in the gc is currently limited , but observation by @xcite and @xcite show the existence of at least some young binaries in the galactic center . in particular , they found that one of the brightest o / wr star in the gc is a massive binary star ( @xmath136 ) .
more recent works @xcite found a few additional binary o - star candidates , though no binaries had been found among the young b - stars .
these studies either rely on detection of eclipsing binaries or through radial velocity detection of short - period binaries , and are therefore mostly sensitive to close binaries .
it is therefore difficult to estimate the underlying binary fraction among the young stars in the galactic center using these limited statistics .
taking them at face value would suggest a very high binary fraction , comparable or even larger than the high binary fraction observed in the filed for similarly young massive stars @xcite . on the theoretical side ,
most binaries in the galactic center are expected to be `` soft '' binaries and therefore have relatively short lifetime due to perturbations by other stars , or merge due to secular perturbations by the massive black hole @xcite .
the binary fraction among the old stellar population in the nuclear cluster is likely to be negligible , and we do not include it in any of our calculations .
binaries in the disk could , however , be `` hard '' in respect to the other disk stars , given the low - velocity dispersion in the disk .
the young age of the disk also suggests that most of the initially formed binaries should still survive .
overall it is therefore likely that the binary fraction of the massive stars in the galactic center stellar disk is high as in the field ( @xmath137 ) , and that the binary fraction does not evolve much since its formation .
as we discussed above , binaries are generally not very important for the disk evolution , but we do consider them in some of our models for completeness , in which case we assume a binary fraction of @xmath138 . a stellar cluster around a massive black hole is expected to evolve into a stellar cusp structure over a relaxation time @xcite . for a spherically symmetric distribution of equal - mass stars analytic considerations and and n - body simulations show that in equilibrium the density profile of the cluster has a power - law distribution , @xmath139 , with @xmath140 @xcite .
mass segregation in multi - mass clusters produce mass - dependent density profiles for the different mass stellar populations , with @xmath141 , where @xmath142 , with @xmath143 the mass of the heaviest stellar element @xcite .
@xcite used fokker - planck calculations to show that @xmath144 should be @xmath145 for solar mass ms stars and up to @xmath146 for stellar black holes ( sbh ) .
similar results from monte - carlo calculations were found by @xcite however , observations of red - giants in the galactic center suggest a core - like structure in the inner regions of the gc ( @xmath144 in the range @xmath147 ; @xcite .
nevertheless , it is not yet clear whether the distribution of red - giants reflect the overall distribution of stars in the gc , and various models had been suggested to explain both a `` real '' core distribution @xcite or an apparent one ( i.e. , only reflecting the distribution of red - giants ; @xcite ) . in light of the above discussion , we consider two possible models : ( 1 ) a cusp model dominated by a steep power - law distribution of sbhs ( @xmath148 ) and ( 2 ) a core model with a shallow power - law distribution ( @xmath149 ) .
in addition to our study of the role of the different nsc and disk components , we also considered the effects of stellar evolution , which become especially important once the long term evolution of the stellar disk is explored .
we considered the evolution of disks with various properties ( see below ) both at times comparable to the observed stellar disk in the gc ( up to 10 myrs ) , as well as a longer term ( 100 myrs ) evolution .
we first considered simple models in which we did not introduce any stellar evolution , and assumed stars do not change over time .
we then consider the effects of stellar evolution in a simplified manner , for each stellar population in a given mass - bin we consider the appropriate ms lifetime for a star of such mass , @xmath150 , according to stellar evolutionary models .
after that period we replace the star in the model with the stellar remnant it produces , assuming a simplified prescription as described in table [ tab : mass_loss ] ; in particular the continuous mass - loss process is simplified and the mass loss is assumed to be immediate .
we only considered evolution of up to 100 myrs and hence we do not consider the stellar evolution of stars with masses smaller than 6 m@xmath151 .
.[tab : mass_loss ] the final mass and final type after @xmath150 ( of each star ) for different intervals of mass
. however star with initial mass less than @xmath152 has @xmath150 greater than 100myr . [
cols="^,^,^",options="header " , ]
in the following we present the evolution of stellar disks for the different models considered .
we first study simpler cases with single - mass disks and then progressively consider more and more realistic cases .
these include disks composed of stars with a range of masses and different mass - functions , and the effects of different types of nscs . in this case
we also explore the differential evolution of stars of different masses in the same disk and their stratification .
we then consider the effects of stellar evolution on the disk , and the long term ( 100 myrs ) evolution of stellar disks .
[ fig : single ] shows the evolution of the rms eccentricity of the disk stars for disks with single - mass stars ( models i , ii , iii and xi ) , for different choices of mass ( 10 and 20 @xmath130 stars ) . as expected the cusp heating has a stronger effect on lower mass stars .
evolution of rms eccentricity of single - mass disk stars .
upper panel shows models i , ii , iii and xi .
bottom panel shows evolution of similar models but with disk stars mass of @xmath153 .
also shown are the measured rms eccentricity of the o - stars in the gc disk@xcite . ]
evolution of rms eccentricity of single - mass disk stars .
upper panel shows models i , ii , iii and xi .
bottom panel shows evolution of similar models but with disk stars mass of @xmath153 .
also shown are the measured rms eccentricity of the o - stars in the gc disk@xcite . ]
[ fig : disk_vs_cusp_core ] shows the role of the cusp and its structure on the evolution of disk stars , and the role of the single - star masses . the final rms eccentricity ( after @xmath154 )
is shown for a range of single - mass disks ( models i , ii and iii ) .
each point in the lines corresponds to a disk with the same total mass , but the mass of the stars composing the disk , indicated on the x - axis ( same for all stars in the disk ) differes .
disks composed of lighter stars can be seen to be more sensitive to the heating by nsc stars , with the cusp model being more effective than the core one .
in contrast , disks composed of heavy stars are cooled down by the nsc ( compared with the model with no existing nsc ) . the rms eccentricity of disk stars after @xmath155 of evolution .
the solid line corresponds to model ii , dot - dashed is for core model iii , and the dashed line corresponds to the case where no nsc exists . ] in the previous section we studied the evolution of stellar disks and their interaction with single - mass nsc stellar population . in the following we consider a more complex and potentially more realistic model for the nsc . in our multi - component
nsc we follow model xi , where the nsc includes both sbhs and ms solar mass stars . fig .
[ fig : multi_cusp ] shows the disk evolution in such model .
though the number of sbhs in the cusp is relatively small , they can play an important role in heating a disk of massive 20 m@xmath151stars , where more massive sbhs lead to stronger heating of the disk .
evolution of a stellar disk composed of 20 m@xmath151 stars , embedded in a multi - component cuspy nsc ( model xi ) .
the left panels shows the comparison between the disk+multi - component cusp model and a model without an nsc component .
the right panel shows the evolution of a disk embedded in cuspy nscs , where different masses for the nsc sbh population are considered .
also shown are the measured rms eccentricity of the gc o - stars @xcite . , title="fig : " ] evolution of a stellar disk composed of 20 m@xmath151 stars , embedded in a multi - component cuspy nsc ( model xi ) .
the left panels shows the comparison between the disk+multi - component cusp model and a model without an nsc component .
the right panel shows the evolution of a disk embedded in cuspy nscs , where different masses for the nsc sbh population are considered .
also shown are the measured rms eccentricity of the gc o - stars @xcite .
, title="fig : " ] on the left panel is plotted the evolution through the time .
the right panel is the difference between the final eccentricity for each model compared to model i.,title="fig : " ] on the left panel is plotted the evolution through the time .
the right panel is the difference between the final eccentricity for each model compared to model i.,title="fig : " ] fig .
[ fig : multi_cusp-1 ] shows that the inclusion of sbhs in addition to solar mass stars in the cusp ( model xi ) leads to a more prominent disk heating , as expected . in models
@xmath156 we studied the evolution of multi - mass stellar populations in a disk , where we considered both disks with salpeter imfs as well as disks with top - heavy imfs . for each model we divided the masses into 20 logarithmic mass bins for the salpeter imf cases ; for better mass resolution
we considered linear mass bins for the top - heavy imf . for each mass bin
we add a coupled equation ( [ eq : dsigmaj_dt ] ) , where we took the mass weighted average to represent the mass bin , with the appropriate number of stars in that mass bin .
[ fig : imf ] shows that the disk evolution leads to a mass stratified structure of the disk , with lighter stellar populations being excited to higher eccentricities / inclinations ( and larger scale height for their disk component ) compared with populations of more massive stars , which are segregated to more circular orbits in the central part of the disk ( lower scale height ) .
such stratification is also observed for models including the effect of the nsc , where only small differences exist as a function of the stellar masses ; lower mass stars are more sensitive to the nsc heating ( not shown ) , as expected .
the top - heavy imf models include a significantly larger number of more massive stars , leading to a much faster heating of the stellar disk , as also observed in the simple models considered by @xcite .
as can be seen in fig .
[ fig : imf_rms ] , the nsc can further contribute to the disk heating , leading to @xmath157 higher rms eccentricity , but its contribution becomes negligible in the top - heavy imf cases , where the heating is dominated by the massive stars in the disk . in order to compare these results with the measured rms eccentricity of the o - stars in the gc we now consider only the rms eccentricity of 3 - 40 m@xmath151 stars , more comparable to the o / b stars observed in the disk today .
for an age of @xmath158 myrs for the gc stellar disk we expect an rms eccentricity of the order of 0.14 , significantly lower than the measured @xmath121 @xcite .
evolution of the rms eccentricity of stars in the mass range @xmath159 corresponding to ob stars currently observed in the gc disk . ] fig .
[ fig : long_term_top ] shows the long term ( 100 myr ) evolution of the stellar populations in disks with continuous ( salpeter or top heavy ) imfs , where simplified stellar evolutionary mass loss is considered ( see section [ sub : stellar - evolution ] ) .
as each of the modeled disks evolve , progressively more massive stars end their life and become lower mass stellar remnants .
these ( now ) lower - mass remnants can then be heated much more effectively through mass - segregation processes .
once the more massive stars evolve the role of the cusp heating becomes more significant , as be seen from the comparison between the no - nsc and cuspy nsc models .
moreover , with the absence of high mass stars intermediate mass populations begin to cool , when no cusp exists ; but these same populations continue to heat up when a cusp population do exist .
in this work we studied the evolution of a stellar disk around a mbh .
we followed previous analytic models developed by @xcite , which considered two - body relaxation and mass segregation processes , and extended the models to include the effect of binary heating , and role of the nuclear cluster interactions with the disk , using several plausible models for the nsc .
we also studied realistic mass function models for the disk stellar populations , and included simplified models for the stellar evolution of the stars .
finally , we also explored the long term ( 100 myrs ) evolution of such disks . in the following we discuss our results and their implications for the role played by the different processes and components in sculpting the disk evolution .
similar to the results of @xcite , we confirm that for isolated stellar disks , the disk heating is dominated by dynamical friction /mass - segregation processes in which the more massive stars heat the low mass stars .
correspondingly , the more massive stars are cooled by the interaction with the low - mass stars , but two - body relaxation of the massive stars by themselves still keeps heating them .
models with top - heavy mass function include a larger fraction of massive stars , and therefore allow for more rapid heating of the disk , producing larger rms eccentricities . our models considered detailed mass - functions for the stars in the disk , allowing us to estimate the evolution of different stellar populations . as seen in fig .
[ fig : imf ] , the disks develop mass stratification where higher mass stellar populations are expected to have significantly lower disk height scale and more circular orbits compared with the low mass stars in the disk .
the existence , or lack of such mass stratification in the gc stellar disk could therefore be an important tool in assessing the processes involved in the evolution of the gc disk and its initial conditions .
the effect of the nsc on the evolution of the stellar disk was not considers before in analytic models .
the nsc stellar population has a high velocity dispersion , i.e. it is an effectively `` hotter '' population than the `` cool '' stellar disk population .
the nsc can therefore potentially heat the stellar disk , even when it is composed of low - mass stars .
our results suggest that a dense cuspy nsc can play a non - negligible , though limited role in heating the stellar disk .
when the disk contains a large population of massive stars , these dominate the disk heating and evolution , and the nsc plays a relatively minor role .
nevertheless , even young disks in which massive stars still exist are affected by the nsc , with top - heavy imf disks being relatively little affects , but salpeter - imf disks showing non - negligible heating up to @xmath157 higher rms eccentricities .
the nsc effects are more prominent in cases where the disk is dominated by low mass stars . these include unrealistic cases of disks composed of low single - mass stars , but they also relevant for the long term evolution of stellar disks with realistic imfs . in the latter case stellar evolution leads to the transformation of massive stars into low mass stellar remnants over time . the effective mass - function of such evolving stellar disks therefore becomes progressively centered around lower mass stars and stellar remnants which replace the previously existing massive stars .
correspondingly , the nsc heating of the cusp plays a progressively more dominant role in the disk evolution as can be seen in fig .
[ fig : long_term_top ] .
in this study we extended the analytic study of stellar disk evolution to include binary heating processes . in principle
the binding energy stored in binaries is significant ; however , the rate of energy exchange between the binaries and the single stars in the disk is slow .
therefore , binary heating is inefficient in heating the stellar disk , compared with two - body relaxation and mass - segregation processes .
these results are consistent with n - body simulation results which included binaries done by @xcite .
note that we consider the overall averaged evolution of the stellar disk population ; atypical single - binary encounters with short - period massive binaries can lead to strong kicks ejecting stars at high velocities , even beyond the escape velocity from the nsc @xcite .
such encounters may produce a small number of outlying stars at highly eccentric / inclined orbits , but are not likely to significantly affect the overall evolution of the disk and its averaged properties ( see also @xcite ) .
we find that the disk evolution over times as long as 100 myrs is slow .
two - body relaxation is too inefficient for the disk to assimilate into the nuclear cluster on such timescales .
the disk structure is expected to keep its coherency , and be observed as a relatively thin disk even at 100 myrs .
this suggests that the existence of older disks formed before the one currently observed in the gc might still be inferred from the stellar kinematics of older , lower mass stars , which may still show a disk - like struture , unless destroyed / smeared by other non - two - body relaxation processes .
our analytic approach considered the role of two - body relaxation and binary heating , and included realistic disk and nsc components . however ,
secular processes and possible collective effects are beyond the scope of such a model ; we briefly review studies of these processes .
resonant relaxation processes @xcite could potentially lead to fast evolution of disk stars inclinations through the vector resonant relaxation ( and slower evolution of the eccentricities ) , especially in the inner parts of the disk @xcite .
@xcite suggested that resonant relaxation can produce significant effects , however they assumed an unrealistic effective mass of nsc stars , which is 5 - 10 times larger than expected in the gc ( even when including stellar black holes ) .
further study of this effect under realistic conditions could shed more light on this issue .
fast two - body relaxation as well as secular evolution of disk stars could also be induced by massive coherent components , massive perturbers @xcite such as molecular clouds , stellar clusters , intermediate black hole or a second stellar disk @xcite .
whether such objects exist / had existed in such configurations as to influence the evolution of the stellar disk is still actively studied .
@xcite considered the collective eccentric disk instability effect .
such effect could lead to rapid change in the distribution of eccentricities in the disk , but it requires initially highly eccentric disk ( see also @xcite ) .
such a process may give rise to a significant population of highly eccentric stars in the disk ; which are not seen in most recent studies @xcite .
it is important to note that by their nature , most of the collective and secular effects mentioned above affect both low mass and high mass stars in a similar manner , leading to similar eccentricity distribution . discerning the kinematic properties of low mass vs. high mass stars is therefore highly important in order to understand the roles of such processes compared with stellar two - body relaxation processes that lead to mass stratification , as shown here in details .
in particular , non - dependence of the stellar kinematics on the stellar masses would strongly suggest collective effects dominate the stellar disk evolution .
the observed eccentricity of o - stars in the stellar disk in the gc , @xmath121 @xcite , is higher than that obtained in the realistic models considered here .
this suggest either initial conditions with significantly higher eccentricities for the disk stars , or that other processes besides two - body relaxation play an important role in the disk evolution .
detailed studies of the eccentricity / inclination distribution of disk stars as a function of mass is not currently available , ( but see @xcite for important global statistical trends ) . moreover ,
lower mass b - stars in these regions may have a different origin than the disk , and may have been captured following a binary disruption @xcite ; such stars would have high eccentricities @xcite and may mask the real distribution of the disk stars eccentricities .
nevertheless , a focused study on stars most likely related to the disk , and their eccentricity - mass relation is highly desirable to asses the existence of possible mass stratification in the stellar disk . finally , two body relaxation processes suggest a factor of two ratio between the eccentricities and inclination of stars in the disk , while different relation might be expected from other processes ( e.g. secular kozai - like evolution induced by an inclined massive structure ) . a good handle on
the eccentricity vs. inclination from observation could give additional clues on the dynamical processes involved .
in this study we analyzed the evolution of a stellar disk around a massive black hole due to two - body relaxation processes and binary heating .
we explored realistic mass functions for the disk stars , and included the effects of two - body relaxation by the nuclear cluster stellar population in which the disk is embedded .
we also considered the effects of stellar evolution and studied the long term evolution of such disks .
the disk evolution is dominated by dynamical friction from the massive stars in the disk , while binary heating plays only a negligible role in the disk evolution .
the nuclear cluster plays a minor role in the disk evolution as long as a large population of massive stars exist , as in the case of a young disk with a top - heavy mass function ; it can play a more significant , though still modest role in heating a disk with a salpeter mass function . due to stellar evolution
the effective mass - function of the disk becomes progressively centered around lower mass stars and stellar remnants which replace the previously existing massive stars .
correspondingly , disk heating by the nuclear cluster plays a progressively more dominant role in the disk evolution .
we find that significant mass stratification arises from mass - segregation processes in the disk .
this could serve as a signature for two - body relaxation dominating the disk evolution .
in contrast , collective and secular effects are typically insensitive to the masses of individual stars in the disk , and would not produce mass stratification .
the observed rms eccentricity of the o - stars in the stellar disk in the galactic center ( @xmath160 ) is larger than obtained in any of our models with realistic conditions , suggesting either an initially hot and/or eccentric disk or that other secular / collective effects play an important role in the disk evolution .
h. , martins f. , fritz t. k. , genzel r. , levin y. , perets h. b. , paumard t. , nayakshin s. , gerhard o. , alexander t. , dodds - eden k. , eisenhauer f. , gillessen s. , mascetti l. , ott t. , perrin g. , pfuhl o. , reid m. j. , rouan d. , sternberg a. , trippe s. , 2009 , , 697 , 1741 h. , martins f. , trippe s. , fritz t. k. , genzel r. , ott t. , eisenhauer f. , gillessen s. , paumard t. , alexander t. , dodds - eden k. , gerhard o. , levin y. , mascetti l. , nayakshin s. , perets h. b. , perrin g. , pfuhl o. , reid m. j. , rouan d. , zilka m. , sternberg a. , 2010 , , 708 , 834 | we study the dynamical evolution of a stellar disk orbiting a massive black hole .
we explore the role of two - body relaxation , mass segregation , stellar evolution and binary heating in affecting the disk evolution , and consider the impact of the nuclear cluster structure and the stellar - disk mass - function .
we use analytic arguments and numerical calculations , and apply them to study the evolution of a stellar disk ( similar to that observed in the galactic center ; gc ) , both on the short ( few myr ) and longer ( 100 myr ) evolutionary timescales .
we find the dominant processes affecting the disk evolution are two - body relaxation and mass segregation where as binary heating have only a little contribution .
massive stars play a dominant role in kinematically heating low mass stars , and driving them to high eccentricities / inclinations .
multi - mass models with realistic mass - functions for the disk stars show the disk structure to be mass stratified , with the most massive stars residing in thinner structures .
stellar evolution plays an important role in decreasing the number of massive stars with time , thereby leading to slower relaxation , where the remnant compact objects of these stars are excited to higher eccentricities / inclinations . at these later evolutionary stages dynamical heating by the nuclear cluster
plays a progressively more important role .
we conclude that the high eccentricities and high inclination observed for the majority on the young o - stars in the galactic center suggest that the disk stars had been formed with initially high eccentricities , or that collective or secular processes ( not explored here ) dominate the disk evolution .
the latter processes are less likely to produce mass stratification in the disk ; detailed study of the mass - dependent kinematic properties of the disk stars could therefore provide a handle on the processes that dominate its evolution .
finally , we find that the disk structure is expected to keep its coherency , and be observed as a relatively thin disk even after 100 myrs ; two - body relaxation is too inefficient for the disk to assimilate into the nuclear cluster on such timescales .
it therefore suggests earlier disks now containing only older , lower mass stars might still be observed in the galactic center , unless destroyed / smeared by other non - two - body relaxation processes .
galaxy : centre galaxies : nuclei galaxy : structure stars : kinematics and dynamics galaxy : nucleus |
although , the double - exchange ( de ) mechanism @xcite could qualitatively explain the phenomenon of colossal magneto - resistance ( cmr ) in manganites , it was found insufficient to provide a consolidated picture accommodating their intricate transport and magnetic properties .
most of the alternate theories proposed recently @xcite , are based on lattice polarons of one type or the other highlighting the strong electron - lattice interactions in these systems .
these polarons and thereby most of the properties of these mn@xmath4 - mn@xmath5 mixed valent compounds depend on the topology of the mno@xmath3 octahedra in their structure . among the recently proposed models , the one based on electronic phase separation ( ps)@xcite has been attracting considerable attention . according the ps model the metal - insulator ( mi ) transitions in these materials
are driven by the percolation of current through ferromagnetic metallic ( fmm ) domains embedded in an antiferromagnetic insulating ( afmi ) matrix .
there have been many structural studies showing the co - existence of such metallic and insulating phases @xcite with sizes varying from nanoscopic to microscopic . the cluster - glass ( cg ) compositions of sm@xmath0ca@xmath1mno@xmath2 system are some of the materials in which such a phase separation could be unambiguously shown using neutron diffraction , owing to the large sizes of the magnetic domains possible in them @xcite .
separation of these magnetic phases in this system has been found to be related to the strong distortions in their gdfeo@xmath2-type structure . also , these distortions control the topology of the mno@xmath3 octahedra of their structure and thereby the one electron band width ( w ) and the electron localization effects . the near fermi level ( e@xmath6 ) electronic behavior of these phase separated systems could possibly be different from those of the la@xmath0sr@xmath1mno@xmath2 which has long been identified as a canonical de cmr system @xcite due to its large w. further , since the term j@xmath7 @xmath8w ( j@xmath7 is the hund s coupling ) which expresses the effective coupling between the @xmath9 and the t@xmath10 electrons of the crystal field split mno@xmath3 octahedra in their structure , remains in the weak - coupling regime , the la@xmath0sr@xmath1mno@xmath2 system is regarded as the least affected by the electron - electron and electron - lattice correlation effects among the cmr manganites . on the other hand ,
the sm@xmath0ca@xmath1mno@xmath2 with its small w and distorted mno@xmath3 octahedra , is more prone to the electron correlation effects such as the localization of charge carriers .
recently , there have been many photoemission studies @xcite highlighting the subtle changes in the near e@xmath6 density of states ( dos ) associated with the mi transitions in phase separated cmr systems .
these shifts in the near e@xmath6 spectral weights lead to charge order ( co ) gaps and pseudogaps . such pseudogaps which have been shown to be a generic behavior of cmr systems with mixed phases@xcite are closely related to the changes in the w and electron localization . as mentioned before
, one expects distinct differences between the mixed phase sm@xmath0ca@xmath1mno@xmath2 and the de driven la@xmath0sr@xmath1mno@xmath2 systems in the nature of their energy gaps across the mi transitions .
ultraviolet photoelectron spectroscopy ( ups ) is a powerful technique capable of probing the fine changes in the valence band electronic structure . in this paper , we show some of the important differences between the sm@xmath0ca@xmath1mno@xmath2 and the la@xmath0sr@xmath1mno@xmath2 systems in their near e@xmath6 electronic structure using high resolution ups .
the doping dependent pseudogap behavior observed in sm@xmath0ca@xmath1mno@xmath2 system is discussed here and compared with that of the de driven la@xmath0sr@xmath1mno@xmath2 .
polycrystalline samples of sm@xmath0ca@xmath1mno@xmath2 and la@xmath0sr@xmath1mno@xmath2 systems were prepared by solid state reactions by mixing mno@xmath11 , cao and sm@xmath11o@xmath2 or la@xmath11o@xmath2 in stoichiometric proportions .
the powders were first heated at 1000 c with intermediate grindings and then pressed in the form of pellets .
they were then sintered at 1500 c for 12 h in air with a slow cooling down to 800 c and finally quenched to room temperature .
details of the sample preparation technique could be found elsewhere @xcite .
the monophasic , homogeneous nature of the samples have been checked by x - ray powder and electron diffraction techniques .
the cationic compositions , close to their nominal values were confirmed using energy dispersive spectroscopy and iodometric titrations .
magnetic and electrical transport properties of the samples were determined using a magnetometer ( equipped with a superconducting quantum interference device ) and four probe resistivity measurements .
a co - existence of two separate phases of fmm and afmi domains below @xmath12 k in @xmath13 and @xmath14 of the sm@xmath0ca@xmath1mno@xmath2 system were shown earlier using neutron diffraction studies carried out at llb , saclay , france on the g41 diffractometer @xcite .
consolidated results of these studies is published eleswhere @xcite .
1.0 cm panel ( a ) : phase diagram of sm@xmath0ca@xmath1mno@xmath2 system .
( b ) : phase diagram of la@xmath0sr@xmath1mno@xmath2 system adapted from the work of chmaissem et al .
@xcite.,title="fig:",width=288 ] angle integrated ultraviolet photoemission measurements were performed using an omicron mu - metal ultra high vacuum system equipped with a high intensity vacuum - ultraviolet source ( his @xmath15 ) and a hemispherical electron energy analyzer ( ea @xmath16 hr ) . at the he @xmath17
( @xmath18 @xmath19 = @xmath20 ev ) line , the photon flux was of the order of @xmath21 photons / sec / steradian with a beam spot of @xmath22 mm diameter .
fermi energies for all measurements were calibrated using a freshly evaporated ag film on a sample holder .
the total energy resolution , estimated from the width of the fermi edge , was about @xmath23 mev for he @xmath17 excitation .
all the photoemission measurements were performed inside the analysis chamber under a base vacuum of @xmath24 @xmath25 @xmath26 @xmath27 mbar .
the polycrystalline samples were repeatedly scraped using a diamond file inside the preparation chamber with a base vacuum of @xmath24 @xmath25 @xmath26 @xmath27 mbar and the spectra were taken within @xmath28 hour , so as to avoid any surface degradation .
all measurements were repeated many times to ensure the reproducibility of the spectra .
for the temperature dependent measurements , the samples were cooled by pumping liquid nitrogen through the sample manipulator fitted with a cryostat .
sample temperatures were measured using a silicon diode sensor touching the bottom of the stainless steel sample plate .
the low temperature photoemission measuements at @xmath29 k were performed immediately after the cleaning the sample surfaces followed by the room tempedrature measurements . in order to make a good comparative study of the two systems ( sm@xmath0ca@xmath1mno@xmath2 and la@xmath0sr@xmath1mno@xmath2 ) , we performed all the measurements under exactly same experimental conditions .
1.0 cm the angle integrated valence band spectrum of the sm@xmath0ca@xmath1mno@xmath2 system taken at @xmath29 k using he i photon energy.,title="fig:",width=288 ]
the phase diagram of the sm@xmath0ca@xmath1mno@xmath2 system published earlier @xcite is shown in fig . 1 ( a ) .
we have chosen five compositions of this system with @xmath30 = @xmath31 , @xmath32 , @xmath33 , @xmath34 and @xmath14 , which are all paramagnetic insulting ( pmi ) at room temperature .
below @xmath12 k , the @xmath35 sample shows a ferromagnetic insulating ( fmi ) behavior .
a large doping region , @xmath36 , shows a co - afmi state at low temperatures .
the cmr properties of these compositions are strongly dominated by this co insulating state .
the low temperature resistivities ( down to @xmath37 k ) of these samples were very high ( @xmath38 @xmath39 cm ) and are thus insulators .
the coexistence of ferromagnetism and metallicity is observed only in the cluster - glass ( cg ) region @xmath40 , @xmath41 , where the resistivity at low temperature ( 10 k ) is @xmath42 to @xmath43 @xmath39 cm , corresponding to that of bad metals .
the cmr effect is observed in the boundary between cg - state ( @xmath44 ) and co - afmi state ( @xmath45 ) .
neutron diffraction studies@xcite have shown that the last two cg compositions ( @xmath30 = @xmath34 and @xmath14 ) have coexisting fmm and afmi phases . fig .
1.(b ) shows the phase diagram of the la@xmath0sr@xmath1mno@xmath2 system , adapted from the work of chmaissem et al.@xcite .
1.0 cm panel ( a ) : high resolution photoemission spectra of the near e@xmath6 region taken at @xmath29 k. a and a refer to the mn 3d e@xmath46 spin up states of the system .
the doping dependent mi transition is accompanied by the shifting of some dos from a to a , referred as the pseudogap in the text .
( b ) : all the other spectra plotted against the one from @xmath30 = @xmath31 depicting the @xmath30 dependent growth of a.,title="fig:",width=288 ] in figure 2 we present the angle integrated valence band spectra of the sm@xmath0ca@xmath1mno@xmath2 samples taken at @xmath29 k. the main features seen in these spectra are by now well - known and looks similar to those reported earlier on different cmr systems @xcite , including the la@xmath0sr@xmath1mno@xmath2 . features seen in the spectra originate from the bonding and antibonding states of mn 3@xmath47 - o 2@xmath48 hybridization .
detailed discussion on these spectral features could be found elsewhere @xcite . as in other transition metal oxide compounds ,
the major contribution to the physical properties of this system comes from the subtle changes in the states near the e@xmath6 ( within 2 ev from the e@xmath6 ) .
a high resolution spectra of this region is shown in fig .
intensities of these spectra are normalized at regions above 1.2 ev and below the e@xmath6 and are shifted along the ordinate axis by a constant value for clarity . here , we concentrate on the subtle changes in the near e@xmath6 spectral features , marked a and a in fig .
different experiments @xcite and band structure calculations @xcite have shown that both these features arise from the mn @xmath49 spin - up states of the crystal field split mno@xmath3 octahedra .
the spectra shown , taken at @xmath29 k , demonstrate the finer shifts in the spectral weight from a to a as we go across the phase diagram shown in fig .
the apparent gaps in the near e@xmath6 dos resulting from these shifts are usually called ` pseudogaps ' .
as mentioned earlier , recent photoemission studies @xcite have highlighted the importance of such pseudogaps , particularly for the models based on electronic phase separation @xcite .
3(a ) shows that some finite number of states build up at a as we go from fmi ( @xmath35 ) to afmi ( @xmath50 ) composition .
further increase of @xmath30 to @xmath33 also results in an increased dos at a. spectra from the @xmath30 = @xmath34 and @xmath14 samples display a distinct feature at a. in panel ( b ) we have plotted the spectra corresponding to each composition together with that from the @xmath30 = @xmath31 for a comparison .
it should be noted that , though all the samples with @xmath30 = @xmath32 , @xmath51 and @xmath33 are afmi at @xmath29 k the number of states at a keep increasing with increase in @xmath30 showing that the pseudogap originating from these shifts exist also in the afmi regime .
this building up of dos at a could be associated with the growing fmm domains in the afmi matrix in the sm@xmath0ca@xmath1mno@xmath2 system with progressive ca doping .
the increase in the dos at a should be due to the shift of states from a to a following the growth of fmm domains .
as we move on to @xmath30 = @xmath34 and @xmath14 , the feature at a becomes quite prominent . in these compositions ,
the fmm domains must be large enough leading to the metallic behavior shown by them .
this has been confirmed earlier using neutron diffraction studies where we @xcite had shown that the sm@xmath0ca@xmath1mno@xmath2 system with @xmath30 = @xmath34 and @xmath30 = @xmath14 have a unique crystalline structure with fmm clusters embedded in a g - type afmi background
. 1.0 cm panel ( a ) : high resolution photoemission spectra of the near e@xmath6 features of la@xmath0sr@xmath1mno@xmath2 compositions taken at @xmath29 k. ( b ) : spectra from different compositions plotted against that from @xmath30 = @xmath52 for comparison.,title="fig:",width=288 ] let us now compare these results from sm@xmath0ca@xmath1mno@xmath2 with those from the canonical @xcite de system , la@xmath0sr@xmath1mno@xmath2 .
figure 4 ( a ) shows the high resolution photoemission spectra of the near e@xmath6 region of la@xmath0sr@xmath1mno@xmath2 system taken at @xmath29 k. these spectra are also normalized and displayed in the same way as described earlier .
the phase diagram @xcite of la@xmath0sr@xmath1mno@xmath2 system ( fig .
1(b ) ) shows that compositions with @xmath30 = @xmath52 to @xmath51 are fmm while @xmath30 = @xmath33 and @xmath53 are afmi .
correspondingly , our spectra from different metallic compositions also show the presence of a fermi edge ; the hallmark of metallicity . in fig .
4 ( b ) we have plotted the spectra from different compositions together with that from the @xmath30 = @xmath52 sample . a close observation and comparison of this figure with fig .
3 ( b ) will reveal that the spectra from different metallic compositions do not show any distinct feature corresponding to a as in the case of the sm@xmath0ca@xmath1mno@xmath2 system . the mi transition in sm@xmath0ca@xmath1mno@xmath2 ( from @xmath30 = @xmath14 to @xmath30 = @xmath31 )
was accompanied by the depletion of the feature at a due to the transfer of some dos from a to a ; namely the pseudogap .
on the other hand , in case of the la@xmath0sr@xmath1mno@xmath2 system we neither find any formation of such a feature at a nor any shift of dos . here
, the mi transition is manifested as the opening up of an insulator gap . the plot in fig .
4 ( b ) , showing the spectra from @xmath54 ( metallic ) together with that from @xmath55 ( insulating ) makes this point clear .
one can see that the insulator to metal transition here is accompanied by an almost uniform increase in the dos over the region between 1.2 ev and the e@xmath6 .
the differences shown by the near e@xmath6 spectral changes associated with the doping dependent mi transitions in the sm@xmath0ca@xmath1mno@xmath2 and the la@xmath0sr@xmath1mno@xmath2 systems could be relevant to the fundamental understanding of the mechanism of cmr .
the insulator - metal transition in the la@xmath0sr@xmath1mno@xmath2 system is due to the hopping of the e@xmath46 electrons from mn@xmath4 to mn@xmath5 via the de and superexchange interactions .
these interactions are sensitive to the distortions in the mno@xmath3 octahedra which control the w and the electron correlation effects . with its small w
these correlation effects are strong in case of the sm@xmath0ca@xmath1mno@xmath2 . as mentioned before , the mixed phase compositions of this system have fmm domains in a co - amfi matrix @xcite . with its strongly distorted mno@xmath3 octahedra
the c - type afmi regions could be more prone to the electron localization effects compared to the fmm domains .
the charge carriers ( e@xmath46 electrons ) of the afmi regions should thereby be more localized compared to the e@xmath46 electrons in fmm regions .
consequently , this can result in higher binding energies for those e@xmath46 electrons in afmi regions compared to the ones in the fmm domains .
as the sizes of these fmm domains increase with ca doping , the number of itinerant e@xmath46 electrons also go up resulting in a shift of spectral weight from a to a.
conversely , this is the origin of the pseudogap . as moreo et al
@xcite have proposed earlier , this pseudogap could be a generic feature of such phase separated systems .
but , the scenario is different in case of the la@xmath0sr@xmath1mno@xmath2 system where such mixed phases do not exist . there
the de and superexchange processes through the undistorted mn - o - mn bond lead only to the hoping of e@xmath46 electrons with no localization processes or pseudogaps involved .
in conclusion , we have shown one of the important differences in the near e@xmath6 electronic behavior of phase separated and de systems using high resolution photoemission experiments .
our study of the doping dependent mi transitions in sm@xmath0ca@xmath1mno@xmath2 compounds shows that the phase separated systems indeed show a pseudogap behavior in their near e@xmath6 electronic spectrum over a large region of their phase diagram .
a comparison of this system with the canonical de la@xmath0sr@xmath1mno@xmath2 system reveals the differences in the nature of the gaps associated with the mi transitions between these two systems .
these differences could be ascribed to the distortions in the mno@xmath3 octahedra of their crystal structures which regulate the electron - electron and electron - lattice correlation effects determining the localization of charge carriers and thereby the pseudogap behavior . | using valence band photoemission we have demonstrated the presence of a pseudogap in the near fermi level electronic spectrum of some of the mixed phase compositions of sm@xmath0ca@xmath1mno@xmath2 system .
the pseudogap was found to grow in size over a large region of the phase diagram of this system , finally leading to a metal - insulator transition .
we have made a study comparing the near fermi level behaviors of this system to those of a canonical double exchange system , namely , la@xmath0sr@xmath1mno@xmath2 .
this study intends to highlight one of the important differences between the phase separated and double exchange colossal magnetoresistance systems in the nature of their energy gaps across the metal - insulator transitions .
these differences could be ascribed to the distortions in the mno@xmath3 octahedra of their structures that regulate the localization of charge carriers .
we have discussed our results from the point of view of models based on the idea of phase separation . |
anteneodo and morgado recently discussed an interesting one - dimensional random walk model that exhibits a phase transition @xcite . in this model , at each time step the random walker moves two lattice spacings to the right or one spacing to the left with probabilities @xmath4 and @xmath5 respectively .
the sites visited by such a walk are , depending on @xmath4 , either part of a single connected cluster of visited sites or many clusters of visited sites ( fragments ) separated by single unvisited sites . a transition from one class to the other takes place at the critical probability @xmath6 . at large times
, for @xmath7 the cluster of visited sites contains no gaps , however , for @xmath8 this cluster contains a finite density of unvisited sites . using monte - carlo simulation data , a power law dependence with an exponent @xmath9
was estimated for @xmath10 , the fraction of fragments with length @xmath2 . since the model is a simple one dimensional random walk that is not known to contain any such anamolous exponents , such a behaviour seems unlikely .
we prove below that the correct asymptotic behaviour of @xmath10 at the critical point is @xmath11^{-1}$ ] which is infact hard to distinguish numerically from a power law dependence with exponent @xmath9 . at large times
, it is easy to see ( using the central limit theorem ) that the probability distribution of the position of the walker is a gaussian with mean @xmath12 and variance @xmath13 .
when the walker is biased towards the left ( @xmath14 ) , the probability of the walker being on the left of the origin tends to @xmath15 as @xmath16 . hence the visited sites are part of a single connected cluster . when the walker is biased towards the right , in general the clusters of visited sites
are separated by single unvisited sites .
the fraction of visited sites @xmath17 is defined as @xmath18 where @xmath19 is the average number of distinct sites visited in an @xmath20-step walk and @xmath21 is the average length of an @xmath20-step walk .
the fraction of unvisited sites @xmath22 , is given by @xmath23 .
for this case , @xmath22 can be calculated exactly using methods outlined in @xcite,@xcite . in the large @xmath20 limit , as @xmath24 , @xmath25 , where @xmath26 .
the total number of fragments in a walk is equal to the number of unvisited sites and hence @xmath27 , the average number of fragments is equal to @xmath28 . the single unvisited sites at the edge of each fragment ensure that the probability of formation of a fragment is independent of the previous history of the walker .
the walk can therefore be considered as a discrete process of independent increments where at each step a new fragment of length @xmath2 is added to the positive edge of the walk with probability @xmath1 .
these probabilities are multiplicative , i.e.- the probability of formation of a fragment of length @xmath29 followed by one of length @xmath30 is @xmath31 .
the number of fragments of length @xmath2 in a given walk is @xmath32 where @xmath33 is the total number of fragments .
therefore @xmath34 and hence @xmath35 since @xmath36 .
for a new fragment of length @xmath2 to be formed the walker must ( i ) be at the positive edge of the walk ( at site @xmath37 for convenience ) , ( ii ) hop over site @xmath38 and reach site @xmath2 by visiting each site between @xmath15 and @xmath2 , without visiting sites @xmath38 and @xmath39 .
thus at the end of each step the walker is once again at the positive edge of the walk .
@xmath1 is therefore the probability of event ( ii ) occuring .
hence @xmath1 is the sum of probabilities of all paths that are consistent with ( ii ) .
we calculate @xmath40 , the probability of condition ( ii ) with the constraint of ` visiting every site between @xmath15 and @xmath2 ' relaxed .
hence @xmath40 is the probability of starting at site @xmath37 and reaching site @xmath2 without visiting @xmath38 and @xmath39 . @xmath40
thus includes the probability of formation of smaller fragments within the segment @xmath41 $ ] . for example @xmath42 . in general @xmath43
this translates to the following equation involving generating functions @xmath44 where @xmath45 and @xmath46 .
+ we calculate @xmath40 by summing over probabilities of all paths that are consistent with the above definition of @xmath40 . for example
, @xmath47 simply involves the walker reaching site @xmath15 from @xmath37 by hopping over site @xmath38 .
thus @xmath48 .
it is also easy to see that @xmath49 is @xmath38 since unvisited sites in this walk can not be separated by two lattice sites . in the calculation of @xmath50
we sum over paths in which the walker starting from @xmath37 , jumps over site @xmath38 to reach @xmath15 and then over @xmath51 to reach site @xmath52 with probability @xmath53 .
once at site @xmath52 , the walker can move two leftward steps followed by a rightward step with probability @xmath54 , any number of times , to return to site @xmath52 .
thus @xmath55 . generalizing this procedure of summing over
all relevant walks , we derive below a closed form expression of @xmath40 in two equivalent ways . the walk can be considered as a markov process where at each time step the walker moves either two lattice spacings to the right or one to the left with probabilities @xmath4 and @xmath5 respectively , independent of the previous step .
the probability @xmath40 can be calculated from @xmath56 , the matrix of transition probabilities @xcite with no transitions from the sites @xmath38 , @xmath39 and @xmath57 . for example
@xmath58\ ] ] where the rows correspond to sites [ 0 1 2 3 4 5 ] .
consider the column vector @xmath59 where @xmath60 , the element in row @xmath61 , is the probability of the walker starting at site @xmath15 , being at site @xmath61 at time step @xmath62 .
now @xmath59 evolves as @xmath63 which ensures that @xmath59 only contains probabilities of paths that never visit sites @xmath38 and @xmath39 .
since the walker starts at site @xmath15 , @xmath64 .
we calculate @xmath65 where @xmath66 is the probability of the walker , starting from site @xmath15 being at site @xmath61 , at any time step .
@xmath65 is thus the sum of @xmath59 at every time step .
hence @xmath67 all walks that start at @xmath15 and end at @xmath2 ( at any time step ) contribute to @xmath40 , therefore @xmath68 where @xmath69 and @xmath70 .
thus @xmath40 is the ( @xmath71 ) matrix element of @xmath72 .
the factor @xmath4 is present because the walker must hop over site @xmath38 to reach @xmath15 as in condition ( ii ) .
we thus obtain @xmath73 , @xmath74 , @xmath75 , @xmath76 and so on . subtituting these into eq.([generating ] )
we obtain @xmath77 , @xmath78 , @xmath79 , @xmath80 and so on .
alternatively , consider @xmath81 , the probability of a walker starting at site -1 , being at site @xmath61 while never visiting @xmath38 and @xmath39 .
now @xmath81 satisfies the difference equation @xmath82 with the boundary conditions @xmath83 and @xmath84 .
a general solution of @xmath81 is @xmath85 which on substitution into eq.([gs ] ) yields the cubic equation @xmath86 .
the solutions to this equation are @xmath87 where @xmath88 . by satisfying boundary conditions we obtain @xmath89 \label{gs - solution}\ ] ] where a is a constant which is determined from the normalization condition @xmath90 . alternatively from the definition of @xmath40 as the connecting probability between @xmath37 and @xmath2 we have @xmath91 we thus have a closed form expression for @xmath40 .
we verify that the two expressions for @xmath40 are equal as the matrix elements of the column corresponding to site 1 in @xmath92 satisfy the same difference equation and , with an appropriate change of variables , the same boundary conditions as g(s ) .
having found the generating function , we can extract the behaviour of @xmath1 for large @xmath2 . from eq.([criticalrho ] ) , @xmath102 where @xmath103 .
we can approximate the summation in the generating function by an integral @xmath104\rho(l)dl$ ] .
the singular part of this integral can be evaluated by equating @xmath105 from which we see that @xmath106 .
alternatively , from the fact that the integral @xmath107/(\log \
l)^2dl$ ] diverges as @xmath108^{-1}(1+o[\frac{1}{\log \epsilon}])$ ] we obtain @xmath109 we thus have an expression for the asymptotic behaviour of @xmath1 .
cases in which the rightward steps are of length @xmath110 can be treated in a similar way . in this case
the independent elements are clusters of visited and unvisited sites separated by @xmath111 zeroes .
i thank prof .
deepak dhar for providing the central ideas that led to these results and r. loganayagam for helpful discussions on the series expansions involved .
+ 9 c. anteneodo and w.a.m .
morgado , phys .
lett . * 99 * , 180602 ( 2007 ) w. feller , _ introduction to probability theory and its applications _ e. w. montroll and g. h. weiss , j. math
* 6 * , 167 ( 1965 ) | we analyse a biased random walk on a 1d lattice with unequal step lengths .
such a walk was recently shown to undergo a phase transition from a state containing a single connected cluster of visited sites to one with several clusters of visited sites ( fragments ) separated by unvisited sites at a critical probability @xmath0 [ _ prl * 99 * , 180602 ( 2007 ) _ ] .
the behaviour of @xmath1 , the probability of formation of fragments of length @xmath2 is analysed .
an exact expression for the generating function of @xmath1 at the critical point is derived .
we prove that the asymptotic behaviour is of the form @xmath3 $ ] . |
the counter - intuitive concept of dissipation - induced generation of quantum states has recently attracted a lot of interest .
it has been proposed to create pure states @xcite and , in particular , entangled states @xcite .
entanglement is among the most striking features of quantum mechanics @xcite and a prerequisite for quantum computation and simulation schemes @xcite . typically , for dissipation to induce entanglement , the dissipative processes require multi - quanta exchange with the environment in addition to carefully engineered coupling constants . realizations of artificial system - environment couplings are , in general , hard to achieve .
on the other hand , it has been shown that multi - quanta dissipation is a natural component in nonlinear systems @xcite .
for instance , graphene - based nanomechanical resonators possess a strong intrinsic nonlinearity @xcite and exhibit nonlinear dissipation @xcite .
in other setups the conservative nonlinearity has to be induced by coupling to auxiliary systems@xcite .
similarly , nonlinear dissipation can be induced in opto - mechanical systems @xcite , and has also been proposed to emerge in superconducting solid state quantum devices @xcite . for a single oscillator mode of a nonlinear system under the influence of two - quanta dissipation , resulting from a nonlinear coupling to the reservoir
, it was shown that nonclassical steady states with non - zero coherences can be generated @xcite .
moreover , the formation of superposition states @xcite as well as non - thermal and squeezed states were shown to be obtainable by two - phonon cooling @xcite . here , we investigate two coupled nonlinear oscillators , which are individually subject to two - quantum dissipation .
we start from a microscopic description of the total system and derive a quantum master equation ( qme ) for the reduced density matrix . by using the rotating wave approximation ( rwa )
we show that in the weak coupling limit , the two - quantum dissipation mainly influences the short - time dynamics .
additionally , dephasing is affecting the system in the long - time limit . by numerically solving the qme and the qme in the rwa
, we demonstrate that at zero temperature entanglement between the oscillators is created , which persists even if the coupling is switched off . for sufficiently large temperatures ,
we find early stage disentanglement ( esd ) , also known as sudden death of entanglement@xcite .
we consider two nonlinear oscillators , which are linearly coupled . additionally , each oscillator is quadratically coupled to a reservoir consisting of harmonic oscillators . accordingly , the total hamiltonian is @xmath0 , where @xmath1 , the masses of the oscillators are @xmath2 , and [ eq : hamiltonian_terms ] @xmath3 here , @xmath4 and @xmath5 denote momentum and oscillation amplitude of oscillator @xmath6 , respectively , and @xmath7 is the anihilation ( creation ) operator of the @xmath6th oscillator .
the oscillators are characterized by their frequencies @xmath8 , the strength of the nonlinearities @xmath9 and the coupling strength @xmath10 . the operator @xmath11 ( @xmath12 ) creates ( destroys ) a phonon in state @xmath13 of reservoir @xmath6 with the frequency @xmath14 .
the coupling strength of oscillator @xmath6 to reservoir state @xmath13 is denoted by @xmath15 . the nonlinear coupling in ( [ eq : hamiltonian_terms4 ] )
will lead to nonlinear damping in the classical limit @xcite . for nanomechanical systems such a coupling
is realized , for example , by coupling of in - plane and flexural motion @xcite ; in opto - mechanical setups @xcite and in superconducting solid state quantum devices @xcite it can be induced by using an auxiliary system . in the weak system - reservoir coupling limit
, the evolution of the reduced density matrix @xmath16 is given by a qme .
following the standard approach by using the born - markov approximation in the interaction picture with respect to @xmath17 one obtains @xcite @xmath18
c_{l m}(\tau ) \nonumber\\ & \:- \big[\ : s_l(t),\rho(t ) s_m(t-\tau ) \big ] c_{m l}(-\tau).\end{aligned}\ ] ] here , the operators @xmath19 and @xmath20 were used to decompose the coupling hamiltonian @xmath21 in the interaction picture . assuming the reservoirs to be initially in thermal equilibrium , @xmath22 , the reservoir correlation functions @xmath23 are given by @xmath24\;,\end{gathered}\ ] ] where @xmath25 is the bose - einstein distribution and @xmath26 is the spectral density .
the specific form of @xmath27 depends on the microscopic details of the system - reservoir coupling .
if @xmath27 is sufficiently smooth around the frequencies of interest , the exact frequency dependence is not crucial . to be specific ,
we use an ohmic spectral density , @xmath28 , where @xmath29 is the nonlinear dissipation strength . in many systems the time - scales associated with @xmath8 and @xmath9
are well separated . in this case
the rwa can be used to simplify the qme . for convenience
we consider a symmetric setup in the following , i.e. , @xmath30 and @xmath31 .
further , we define the one - sided fourier transform of the reservoir correlation function @xmath32 the rates @xmath33 determine the strength of dissipation , while @xmath34 renormalizes the system hamiltonian . using
the expression of the bath correlation function one finds that @xmath35\;,\\ \gamma_m(-2\omega_0 ) = { } & \gamma_0 n(2\omega_0)\;.\end{aligned}\ ] ] in the weak coupling limit , @xmath36 , one finally obtains the qme in rwa @xmath37 + \sum_m \left ( \gamma_m(2\omega_0 ) \mathcal{l}[a^\dag_m a^\dag_m ] \right.\notag\\ { } & \left .
+ \gamma_m(-2\omega_0 ) \mathcal{l}[a_m a_m ] \right ) \rho + \mathcal{d}_{12}(\lambda ) \rho\end{aligned}\ ] ] in schrdinger representation .
the oscillator hamiltonian in rwa is given by @xmath38 and the superoperator @xmath39 is defined as @xmath40\rho = -\frac{1}{2 } x x^\dag \rho - \frac{1}{2 } \rho x x^\dag + x^\dag \rho x\;.\ ] ] thus , to lowest order in @xmath41 , the oscillators are individually coupled to their respective reservoirs .
the other superoperator @xmath42 becomes @xmath43\rho \notag\\ { } & -\frac{1}{2}\upsilon_{- } \big [ ( n_1 - n_2 ) ( a^\dag_1 a_2 - a^\dag_2 a_1 ) \rho \big.\notag\\ { } & - ( a^\dag_1 a_2 - a^\dag_2 a_1 ) \rho ( n_1 - n_2 ) \notag\\ { } & + \rho ( a^\dag_1 a_2 - a^\dag_2 a_1)^\dag ( n_1 - n_2)\notag\\ { } & \big.- ( n_1 - n_2 ) \rho ( a^\dag_1 a_2 - a^\dag_2 a_1)^\dag \big]\ ; , \label{eq : cross_d}\end{aligned}\ ] ] where @xmath44 with @xmath45 $ ] and @xmath46 .
the terms in the qme which are proportional to @xmath47 describe the loss of two quanta into the bath , while the terms which are proportional to @xmath48 give rise to the absorption of two quanta from the bath . for zero temperature
only the former processes are present , since @xmath49 .
in contrast to the contributions involving @xmath50 , the superoperator @xmath51 contains two dephasing terms , proportional to @xmath52 and @xmath53 , respectively .
one sees that these two terms have a different temperature dependence . for @xmath54 , @xmath55 , while for @xmath56 one has @xmath57 .
note that , if the oscillators were linearly coupled to the reservoirs , the latter dephasing terms would only arise if @xmath58 @xcite .
first we concentrate on @xmath59 , which implies @xmath60 .
we introduce the basis vector @xmath61 , which denotes a state with @xmath62 quanta in oscillator @xmath63 and @xmath64 quanta in oscillator @xmath65 . from eq .
one sees that a density matrix , which involves the states @xmath66 , @xmath67 and @xmath68 will not be affected by the dissipation and the qme does not lead to a coupling to other states . in other words , the steady - state will , in general , not be the ground state @xmath66 , but rather a mix of superpositions of @xmath66 , @xmath67 and @xmath68 .
this state can be written as @xmath69 where weights of the respective states are determined by the initial state and evolve according to the hamiltonian @xmath70 and the superoperator @xmath51 .
for example , the populations @xmath71 and @xmath72 can be obtained from the sum of the populations of the even ( @xmath73 ) and odd ( @xmath74 ) states in the initial density matrix and @xmath75 .
] , respectively .
this is a result of the invariance of the total hamiltonian to changes of total parity ( @xmath76 and @xmath77 ) .
a similar behavior is observed for a single oscillator @xcite , where it leads to the formation of a nonclassical state with non - vanishing coherences . to investigate the dynamics of the coupled oscillators the qme and the qme in rwa are solved numerically .
the hilbert space is truncated after @xmath78 states for each oscillator .
equation is solved in the eigenbasis of the system hamiltonian , which corresponds to the wangsness - bloch - redfield method @xcite .
initially , the system is prepared in a product state of two coherent states , @xmath79 , where @xmath80 , @xmath81 is the amplitude of initial displacement , and the rest of the system parameters are @xmath82 , @xmath83 and @xmath84 .
the dynamics of the populations of the states @xmath66 , @xmath68 and @xmath67 is shown in fig .
[ fig : zerot_population](a ) for the situation where only one oscillator is initially displaced ( @xmath85 and @xmath86 ) . after a transient behavior up to @xmath87 ,
the population of the ground state settles to a constant value @xmath88 and oscillations of @xmath89 and @xmath90 can be seen , while @xmath91 remains constant . in the long - time limit , @xmath92 due to the presence of dephasing .
this is shown in fig .
[ fig : zerot_population](b ) .
the transient behavior corresponds to the individual relaxation of each oscillator from the initial state to a state which has at most one excitation per oscillator . using the qme in rwa
the equations of motion for the matrix - elements in the steady - state can be solved .
it is convenient to introduce @xmath93 and the components of the bloch vector @xmath94 , @xmath95 and @xmath96 . in the limit @xmath97 , their solutions are [ eq : anasol ] @xmath98 , \\
u(t ) = & p_{\rm odd}(e^{-2\upsilon t }
-1 ) + u_0 e^{-2\upsilon t } , \\
v(t ) = & e^{-\upsilon t}[v_0 \cos(\lambda t)+w_0\sin(\lambda t ) ] \;.\end{aligned}\ ] ] in the long time limit , @xmath99 and @xmath100 while @xmath101 , due to dephasing . during its time - evolution ,
the bloch vector traces the surface of a cone as can be seen in fig .
[ fig : zerot_population](c ) . without dephasing ( @xmath102 )
it describes a circle . and @xmath86 and ( c , d ) @xmath85 and @xmath103 .
the rest of the parameters are chosen as in fig . [
fig : zerot_population ] . symbols and full lines denote numerical results from the redfield approach and the rwa , respectively .
the dashed lines indicate the asymptotic negativity.[fig : zerot_negativity],title="fig:",scaledwidth=48.0% ] + and @xmath86 and ( c , d ) @xmath85 and @xmath103 .
the rest of the parameters are chosen as in fig .
[ fig : zerot_population ] . symbols and full lines denote numerical results from the redfield approach and the rwa , respectively .
the dashed lines indicate the asymptotic negativity.[fig : zerot_negativity],title="fig:",scaledwidth=48.0% ] to quantify the entanglement of the oscillators , which is created by coupling the two systems , the negativity @xmath104 is computed for each time - step @xcite .
the matrix @xmath105 is the partial transpose of the bipartite mixed state @xmath16 with respect to oscillator @xmath63 .
the negativity corresponds to the absolute value of the sum of negative eigenvalues of @xmath105 and vanishes for any separable state . for the steady state @xmath106 the negativity
is found from the negative roots of the characteristic polynomial of the state s partially transposed density matrix . according to vieta s formula ,
the product of the four roots is @xmath107 this implies that at least one root @xmath108 has to be negative if @xmath89 , @xmath90 and @xmath109 are non - vanishing , which results in a finite negativity . the numerically obtained behavior of @xmath110 for the case with @xmath85 and @xmath86 is shown in figs . [ fig : zerot_negativity](a ) and ( b ) .
an almost periodic behavior is observed , with a vanishing negativity at multiples of a half - period @xmath111 . comparing with fig .
[ fig : zerot_population](a ) one finds that these times correspond to maximal population in either @xmath68 or @xmath67 , which happens twice per period . at those times the coherence between @xmath68 and @xmath67 vanishes . on the other hand ,
maximal negativity is found when the coherence is maximal and @xmath112 , which also happens twice per period .
it is important to realize that the entanglement persists even if the the coupling is switched off , since the steady state is not affected by the nonlinear dissipation , and the dephasing depends on the presence of the coupling @xmath113 .
the time - evolution of the negativity is mainly governed by the dynamics of @xmath114 .
the fast oscillations in fig .
[ fig : zerot_negativity](b ) can be understood by the circular revolution of the bloch vector around the @xmath115-axis in fig .
[ fig : zerot_population](c ) .
the negativity dip can in the same manner be understood in terms of @xmath116 initially being in the vicinity of the phase space origin in addition to the requirement of eq .
being fulfilled .
the negativity saturation is due to @xmath116 reaching its final value at the tip of the cone as @xmath117 . in the case of two displaced oscillators , @xmath118
, the populations @xmath119 and @xmath89 again undergo a short transient behavior and quickly saturate at finite , steady values .
the time evolution of the negativity is shown in figs .
[ fig : zerot_negativity](c ) and ( d ) .
initially , the negativity has decaying oscillations , with a period @xmath120 , which occur due to the influence of decaying coherences between states with more than one excitation . for finite @xmath121 ,
the negativity first decays and then increases to settle at a finite value .
this behavior is governed by @xmath122 , which is initially positive and monotonically decreases according to eq . .
since @xmath123 the bloch vector only points along the @xmath115-axis .
when it crosses the origin , the negativity is minimal .
this is also the reason for the absence of long - term oscillations in fig .
[ fig : zerot_negativity](d ) .
the asymptotic value of the negativity is given by @xmath124 which only depends on the initial state via @xmath74 .
if @xmath121 were zero , @xmath125 , and the negativity would quickly saturate at a finite value , as seen in fig .
[ fig : zerot_negativity](c ) . .
the rest of the parameters are chosen as in fig .
[ fig : zerot_population ] .
the curves are shifted by @xmath126 for convenience .
the temperatures in units of @xmath127 are given by the labels next to the curves .
[ fig : fint_negativity],scaledwidth=48.0% ] at finite temperatures the thermal excitations created by the bath will , in general , lead to a decay of the coherences between the fock states .
therefore , one expects to observe a decay of the entanglement with increasing temperature . in fig .
[ fig : fint_negativity ] the time and temperature dependence of the negativity is shown . for temperatures @xmath128
the negativity is slowly decaying with time . for larger temperatures
the negativity is seen to be zero after a finite time .
this behavior is known as esd @xcite .
these results are also consistent with the previous result @xcite , that the occurrence of esd and of entanglement sudden ( re)birth at finite temperatures are generic features .
in this work we have investigated the quantum dynamics of two weakly - interacting anharmonic oscillators , which are nonlinearly coupled to individual dissipative environments .
this scenario leads to the formation of non - classical steady states and , in particular , entanglement .
additionally , at finite temperatures the exotic features of entanglement sudden death and ( re)birth are observed .
our results show that dissipation - induced quantum state generation is feasible without engineering the system - environment coupling . utilizing the natural presence of nonlinear dissipation in nonlinear nano - scale systems provides a promising route to realize exotic quantum state generation . | the quantum dynamics of two weakly coupled nonlinear oscillators is analytically and numerically investigated in the context of nonlinear dissipation .
the latter facilitates the creation and preservation of non - classical steady states .
starting from a microscopic description of two oscillators individually interacting with their dissipative environments , it is found that in addition to energy relaxation , dephasing arises due to the mutual coupling . using the negativity as an entanglement measure
, it is shown that the coupling entangles the oscillators in the long - time limit . for finite temperatures , entanglement sudden death and rebirth are observed . |
_ * conventions * : we use superscripts to denote real dimension e.g. a manifold denoted @xmath4 has dimension @xmath5 .
once the dimension is established we omit the superscript so that the manifold @xmath4 is also denoted @xmath6 . a group @xmath7 is said to _ split over a subgroup @xmath8 _ if @xmath7 has one of the following descriptions : + case i ( free product with amalgamation ) @xmath9 with @xmath10 , + case ii ( hnn extension ) @xmath11 , @xmath12 . _
the presence of a group action on a space often allows one to promote an existing structure to another of a more strictly controlled kind .
examples of this phenomenon in topology include papakyriokopoulos s sphere theorem @xcite , waldhausen s torus theorem @xcite , and the geometric superrigidity theorem of mok _ et al _ , @xcite . in this paper
we propose the following topological result which has features in common with them all .
[ main ] let @xmath13 be a closed , orientable , aspherical topological manifold with @xmath14 even and @xmath15 , satisfying the following properties : 1 .
[ fbox ] every cellular action of @xmath16 on a cat(0 ) cubical complex has a global fixed point , 2 .
[ k_0 ] the projective class group @xmath17 vanishes for any torsion free finite extension @xmath18 of @xmath16 . given
any closed , orientable , aspherical topological manifold @xmath19 and any @xmath2-injective , continuous function @xmath20 , there is a diagram as follows which commutes up to homotopy [ diagram1 ] @xmath21 & n \ar@{->}_j[dd ] \ar@{<-->}[l]_{{h ' } } \\ n ' \ar@{-->}_i[d ] & \\m ' \ar@{<-->}[r]^{h } & m } $ ] where : 1 .
@xmath22 are closed , orientable , aspherical , topological manifolds and @xmath23 are homotopy equivalences , 2 .
the map @xmath24 is a finite degree cover , and 3 .
@xmath25 is a two sided topological embedding .
the reduced projective class group is conjectured to vanish for all torsion free groups ( see for a discussion ) so condition ( 2 ) of theorem is conjecturally unnecessary .
condition ( 1 ) on the other hand is at the heart of the result . by sageev
s duality theorem @xcite together with scott s ends theorem @xcite it is equivalent to the assertion that no covering space of @xmath6 has more than 1 topological end .
this is central to the proof which proceeds in 2 stages , first to derive a splitting of the fundamental group of @xmath26 as an hnn extension or as a non - trivial amalgamated free product and then to use the splitting to obtain the required decomposition of @xmath26 .
the first stage is essentially group theoretic and requires condition 1 , while the second uses surgery theoretic techniques which necessitates the restriction to dimensions @xmath27 .
the restriction to even dimensions is unnecessary ; however , the surgery methods required for the odd dimensional case are more complicated and we address them in a forthcoming paper .
subject to the borel conjecture , there is a more elegant formulation of theorem [ main ] .
the borel conjecture , formulated by armand borel in 1953 asserts that any homotopy equivalence between aspherical manifolds should be homotopic to a homeomorphism . since the homotopy type and the dimension of an aspherical manifold is determined by its fundamental group
the borel conjecture asserts that the fundamental group is sufficient to classify aspherical manifolds of any given dimension .
if in theorem [ main ] , both @xmath28 and @xmath16 are known to satisfy the borel conjecture , or , alternatively , if @xmath16 satisfies the borel conjecture and is square root closed in @xmath28 , then the homotopy equivalences @xmath29 are homotopic to homeomorphisms and the commutative diagram simplifies as below . in this case
the theorem asserts that every @xmath2-injective codimension-@xmath30 map @xmath31 factors , up to homotopy , as a finite cover of a @xmath32-sided embedding .
[ diagram2 ] @xmath33 \ar@{-->}[ld]_{p\circ h ' } \\ n ' \ar@{^{(}-->}[rd]_{i\circ h } & \\ & m } $ ] [ superrigid ] the hypotheses on @xmath6 are rigidity constraints which , for example , are satisfied when @xmath6 is a closed , orientable , riemannian manifold whose universal cover @xmath34 is quaternionic hyperbolic or the cayley hyperbolic plane . more generally , by @xcite , condition [ fbox ] is satisfied whenever @xmath16 satisfies kazhdan s property ( t ) .
by results of bartels and lck @xcite , condition [ k_0 ] is satisfied whenever @xmath16 is a word hyperbolic group . indeed ,
any finite extension of a hyperbolic group is itself , hyperbolic and so @xcite applies . for
the following reasons condition [ k_0 ] is also satisfied when @xmath16 is cat(0 ) , i.e. _ it admits a co - compact isometric proper action on a finite dimensional cat(0 ) space_. assume that @xmath35 admits a co - compact , proper isometric action on a cat(0 ) space @xmath36 and that @xmath18 is a finite extension of @xmath8 . letting @xmath37 denote the intersection of the conjugates of @xmath8 in @xmath18 we exhibit @xmath18 as a finite extension @xmath38 by the universal embedding theorem ( * ? ? ?
* theorem 2.6a ) , @xmath18 embeds in the standard wreath product @xmath39 which acts cocompactly and properly discontinuously by isometries on the the @xmath40-fold direct product @xmath41 of @xmath36 equipped with the induced cat(0 ) metric .
hence , by @xcite the farrell - jones conjecture holds for the wreath product @xmath42 .
however , as the property of a group satisfying the farrell - jones conjecture is preserved under taking subgroups , we conclude that @xmath18 satisfies the farrell - jones conjecture .
consequently the projective class group vanishes for every torsion - free finite extension of @xmath16 .
hence we obtain as a corollary : [ superigid ] let @xmath19 and @xmath43 be closed , aspherical , orientable topological manifolds with @xmath14 even and @xmath15 such that @xmath16 is either a word - hyperbolic or a cat(0 ) group that satisfies kazhdan s property ( t ) . if @xmath26 satisfies the borel conjecture then every @xmath2-injective map @xmath31 is , a finite cover of an embedding ( up to homotopy ) .
we regard this as a topological counterpart to the celebrated geometric superrigidity theorem : [ superrigidity ] let @xmath34 be a globally symmetric irreducible riemann manifold of non - compact type .
assume that either @xmath34 is of rank at least @xmath32 , or is the quaternionic hyperbolic space of dimension at least @xmath44 or the hyperbolic cayley plane .
let @xmath8 be a cocompact discrete subgroup of the group of isometries of @xmath34 acting freely .
let @xmath45 be a riemann manifold .
let @xmath46 be a non - constant @xmath8-equivariant harmonic map from @xmath34 to @xmath45 .
when the rank of @xmath34 is at least @xmath32 , the riemann sectional curvature is assumed to be non - positive . when the rank of @xmath34 is one , the complexified sectional curvature is assumed to be nonpositive .
then the covariant derivative of the differential of @xmath46 is identically zero . as a consequence ,
@xmath46 is a totally geodesic isometric embedding ( up to a renormalization constant ) . to compare the two results note that the map @xmath31 induces a @xmath16 -equivariant map between the universal covers .
the constraint on the curvature of the target @xmath47 in the geometric super rigidity theorem is dropped ( together with the requirement that the target is smooth ) in favour of a statement that @xmath47 is contractible and @xmath28 satisfies the borel conjecture . instead of a harmonic map , we start with a continuous function which is codimension-1 .
the conclusion that the map is totally geodesic up to renormalisation is replaced by the conclusion that it is a finite cover of an embedding up to homotopy .
deforming the map @xmath48 to a harmonic map in the same homotopy class allows us to combine the conclusions of the two results to see that the embedding provided by theorem [ main ] is homotopic to a totally geodesic surface in @xmath26
. the reader may find it helpful in visualising the results in this paper to consider the analogous statements in lower dimensions .
first we consider the case of @xmath2-injective loops on a @xmath32-torus @xmath49 .
the fundamental group @xmath50 is free abelian of rank @xmath32 and so any homotopy class of curves is represented by a pair of integers @xmath51 .
it is an elementary fact that a non - trivial curve is homotopic to a simple closed curve if and only if the pair @xmath51 is coprime , and it follows that in general a curve representing a pair @xmath52 is homotopic to a degree @xmath53 cover of the embedded curve representing the pair @xmath54 where @xmath55 .
it follows that every closed @xmath2-injective curve is homotopic to a finite cover of an embedded loop .
this is a direct analogue of the codimension-@xmath30 topological rigidity theorem .
now , in contrast , consider the case of loops on a _ hyperbolic _ surface .
every orientable hyperbolic surface @xmath56 admits @xmath2-injective maps @xmath57 which do not factorise up to homotopy as a finite cover of an embedding .
recall that the free homotopy class of a closed loop contains a unique geodesic , and that this minimises the self intersection number for curves in that class . on the other hand ,
intersection numbers are multiplicative on powers , and it follows that any self intersecting closed geodesic which has intersection number @xmath30 with some simple closed curve provides a loop which does not finitely cover an embedded loop so the analogous statement fails in this case .
such surfaces do carry many splitting curves , and these can often be obtained from immersed totally geodesic curves by the somewhat different methods of cut and paste . in dimension
@xmath58 the situation worsens : the kahn and markovic theorem , @xcite , shows that every hyperbolic @xmath58-manifold contains an immersed @xmath2-injective surface , however , there are examples which do not contain any embedded @xmath2-injective surfaces which they could cover .
for example , in @xcite , reid constructs a non - haken manifold @xmath59 which admits a finite cover homeomorphic to a hyperbolic surface bundle over @xmath60 .
the fibre in the finite cover yields an immersed surface in @xmath26 but there are no embedded surfaces in @xmath26 which it could cover so there is in general no analogue of the codimension-@xmath30 topological rigidity theorem in low dimensions .
this leaves open the question of what happens in dimension @xmath61 .
the strategy in the proof of theorem [ main ] is to first establish the existence of a splitting of the fundamental group @xmath28 over a suitable subgroup using methods from geometric group theory .
this result applies in the context of poincar duality groups and is of independent interest .
[ pd ] let @xmath7 be an orientable @xmath62 group and @xmath8 be an orientable @xmath63 subgroup of @xmath7 .
if every action of @xmath8 on a cat(0 ) cube complex has a global fixed point ( in particular , by @xcite , if @xmath8 satisfies kazhdan s property ( t ) ) , then @xmath7 splits over a subgroup @xmath18 containing @xmath8 as a finite index subgroup .
theorem [ pd ] may be viewed as an analogue of the algebraic annulus and torus theorems of kropholler and roller @xcite , dunwoody and swenson @xcite . in a companion paper
we give a group theoretic application of theorem [ pd ] , obtaining a canonical decomposition of poincar duality groups over codimension-1 property ( t ) poincar duality subgroups .
this may be viewed as an analogue of the algebraic jsj decomposition studied by kropholler @xcite , dunwoody and sageev @xcite , scott and swarup @xcite and others .
while theorem [ main ] is stated in the topological category the surgery technology applied in this paper also works in the smooth category and if the map @xmath48 is smooth then the map @xmath64 it constructs is smooth also .
we will use this fact to deduce the following obstruction result from theorem [ main ] .
[ firstcorollary ] let @xmath65 be a closed , orientable , aspherical , smooth manifold such that @xmath66 and the first betti number @xmath67 is zero .
let @xmath68 be a closed , orientable , aspherical , smooth manifold with at least one non - zero pontryagin number such that @xmath16 is either word hyperbolic or a cat(0 ) group and satisfies kazhdan s property ( t ) .
then there are no @xmath2-injective continuous maps @xmath69 .
the proof of theorem [ main ] relies on ideas from geometric group theory , surgery theory , homological algebra and rigidity theory and the paper is organised as follows . in section [ prelim1 ] we set up and prove the algebraic splitting theorem , theorem [ pd ] , for poincar duality groups . in section [ prelim2 ]
we outline the topological ingredients required for the proof of the topological splitting theorem , theorem [ main ] . in section [ proof ]
we present the proof of theorem [ main ] , and in section [ application ] we give the proof of the obstruction result , corollary [ firstcorollary ] .
in order to set up notation we will first define poincar duality groups . for further details ,
we refer the reader to @xcite and references therein .
let @xmath70 be a commutative ring with unity .
a group @xmath7 is said to be a duality group of dimension @xmath14 over @xmath70 if there is an @xmath71 module @xmath72 such that for all @xmath73 and for all @xmath71 modules @xmath74 , one has the following natural isomorphisms ( referred to as duality isomorphisms ) . @xmath75 here , @xmath7 acts diagonally on the tensor product .
the module @xmath72 is called the dualising module of @xmath7 .
it is clear that the cohomological dimension over @xmath70 of such a group is at most @xmath14 .
moreover , taking the module @xmath74 to be the induced module @xmath71 in the duality isomorphism and applying the eckmann - shapiro lemma , we find that @xmath76 for all @xmath77 and @xmath78 .
this implies that @xmath7 is a group of type fp over @xmath70 ; however , much more is true as the following theorem shows .
* theorem 9.2 ) a group @xmath7 is a duality group of dimension @xmath14 over @xmath70 if and only if the following three conditions hold . 1 .
@xmath7 is of type fp over @xmath70 2 .
@xmath79 for @xmath77 3 .
@xmath80 is flat as an @xmath70-module .
a group @xmath7 is called a poincar duality group of dimension @xmath14 over @xmath70 ( or a @xmath81-group , for short ) if @xmath7 is a duality group of dimension @xmath14 over @xmath70 and the dualising module @xmath72 is isomorphic to @xmath70 . when @xmath7 is the fundamental group of a closed aspherical manifold @xmath26 , then poincar duality holds with @xmath82 and in this case the dualising module is a trivial @xmath7-module if and only if @xmath26 is orientable .
we will follow the convention of writing @xmath83 for @xmath84 . in this paper
we will consider @xmath83 groups and @xmath85 groups .
observe that every @xmath83 group is also a @xmath85 group .
a group is a one - dimensional duality group over @xmath86 if and only if it is finitely generated free .
consequently , a group @xmath7 is a @xmath87 group if and only if @xmath88 . that every @xmath89 group is a surface group is a deep theorem due to bieri , eckmann , linnell and muller .
it is conjectured that every finitely presented @xmath83 group is the fundamental group of a closed aspherical manifold .
the geometric and algebraic end invariants which play a crucial role in this paper may be defined in terms of the cohomology groups of a group with exotic coefficients . in order to define them
we will need the following modules .
let @xmath90 denote the collection of all subsets of @xmath7 .
then , @xmath90 is an @xmath91-vector space with respect to the operation of symmetric difference .
one checks that @xmath90 is also a @xmath7 module .
moreover , @xmath92 @xmath93 . for a subgroup @xmath94
we denote by @xmath95 the @xmath96-module @xmath97 : @xmath98 similarly the power set @xmath99 of @xmath100 and the collection of finite subsets of @xmath101 , written @xmath102 are @xmath103 $ ] modules .
in fact , @xmath99 @xmath104 and @xmath105 @xmath106 .
we recall the following definitions from @xcite : for a subgroup @xmath94 the algebraic end invariant is defined as @xmath107 and the geometric end invariant is defined as @xmath108 if @xmath8 has finite index in @xmath7 , then coinduction coincides with induction and therefore @xmath109 .
if on the other hand , @xmath8 is of infinite index , then one can study the long exact sequence in cohomology corresponding to @xmath110 and an easy computation yields the formula @xmath111 we now have all the notation required to state our splitting theorem for @xmath83 groups .
[ dualitytheorem ] let @xmath7 be an orientable @xmath62 group and @xmath8 be an orientable @xmath63 subgroup of @xmath7 .
if every action of @xmath8 on a cat(0 ) cube complex has a global fixed point , in particular , if @xmath8 satisfies kazhdan s property ( t ) , then @xmath7 splits over a poincar duality subgroup @xmath18 containing @xmath8 as a finite index subgroup .
let @xmath7 and @xmath8 be as in the statement of the theorem . as @xmath112 groups are also @xmath113 groups , we may work over @xmath91 .
since the dualising module @xmath114 is trivial the duality isomorphism gives @xmath115 for all @xmath73 .
therefore , @xmath116 where the last isomorphism is given by the classical eckmann - shapiro lemma .
using duality isomorphisms for @xmath8 , we get @xmath117 . as @xmath118
, we deduce that @xmath119 .
we now invoke lemma 2.5 of @xcite to get a subgroup @xmath120 of at most index @xmath32 in @xmath8 such that @xmath121= @xmath122=2 .
applying sageev s construction ( * ? ? ?
* theorem 2.3 ) we obtain a @xmath123 cube complex @xmath36 such that @xmath7 acts essentially on @xmath36 and @xmath120 is the stabiliser of an oriented codimension-1 hyperplane @xmath124 . as @xmath120 has finite index in the group @xmath8 the action of @xmath120 on the cat(0 ) cube complex @xmath124 has a global fixed point .
one now extracts from the fixed point of the action , a proper @xmath120 almost invariant subset @xmath125 of @xmath7
such that @xmath126 ( * ? ? ?
* lemma 2.5 ) .
the singularity obstruction @xmath127 , introduced in @xcite , is defined as the collection @xmath128 , where @xmath129 @xmath130 .
we now apply as follows to deduce that the subgroup @xmath131 satisfies @xmath132 . *
when @xmath133 then @xmath134 is a @xmath135-almost invariant subset .
* when @xmath136 then @xmath137 is a @xmath135-almost invariant subset .
* when @xmath138 then @xmath139 is a @xmath135-almost invariant subset .
* when @xmath138 then @xmath140 is a @xmath135-almost invariant subset .
we will use this to show that the elements of @xmath127 all lie in the commensurator of @xmath120 , allowing us to apply the generalised stallings theorem from @xcite .
* claim * for any @xmath85 group @xmath141 with subgroup @xmath142 , if @xmath143 then @xmath144 @xmath145 .
let @xmath146 be as in claim . if @xmath147 , then @xmath142 has infinite index in @xmath141 and so by strebel s theorem @xmath148 .
we will show that in fact we have equality .
suppose not and @xmath144 @xmath149 .
as before @xmath147 implies that @xmath150 is non - zero .
however , by duality for @xmath141 and the eckmann - shapiro lemma , we have @xmath150 @xmath151 @xmath152 . as @xmath144 @xmath149 , there is a projective resolution @xmath153 of @xmath91 by @xmath154-modules of length @xmath155 . by definition
, @xmath156 is the @xmath157-th homology of the complex @xmath158 . clearly , the latter vanishes , contradicting the fact that @xmath150 is non - zero .
this proves the claim .
returning to the proof of theorem [ dualitytheorem ] we deduce from the claim that if @xmath159 then @xmath135 has the same cohomological dimension as @xmath120 and as @xmath160 .
it follows , again by strebel s theorem that @xmath135 has finite index in both .
hence , @xmath161 lies in the commensurator @xmath162 of @xmath120 as required .
therefore by theorem b of @xcite , @xmath7 splits over a subgroup @xmath18 commensurable with @xmath120 and hence with @xmath8 .
more precisely , @xmath7 is either a non - trivial amalgamated free product @xmath163 or an hnn extension @xmath164 such that @xmath18 and @xmath8 are commensurable .
now consider the action of @xmath7 on the bass - serre tree for the splitting .
since @xmath8 is commensurable with an edge stabiliser , and @xmath7 acts with no edge inversions , the subgroup @xmath8 fixes a vertex @xmath165 so up to conjugation ( and switching the roles of @xmath166 if necessary in the amalgamated free product case ) we may assume that @xmath8 lies in the vertex stabiliser @xmath167 .
having taken a conjugate of the picture to arrange for @xmath168 it is still the case that @xmath8 is commensurable with an edge stabiliser @xmath169 for some @xmath161 , but , _ a priori _ it is not clear that @xmath8 is commensurable with @xmath18 itself .
suppose that @xmath169 fixes an edge @xmath170 so that @xmath171 fixes both @xmath170 and @xmath165 .
it then fixes the first edge @xmath172 on the geodesic from @xmath165 to @xmath170 .
so the stabiliser of this edge is commensurable with @xmath8 . up to conjugation within @xmath167
this stabiliser is @xmath18 , so we may express @xmath7 as an amalgamated free product @xmath163 or as an hnn extension @xmath164 with @xmath168 and so that @xmath8 and @xmath18 are commensurable . by (
* theorem v.8.2 ) we obtain a @xmath83 pair @xmath173 , where @xmath174 is the set of cosets of @xmath18 in @xmath167 .
there are two types of @xmath83 pairs , the @xmath175-bundle type where @xmath176 has two elements and the general type where @xmath174 is infinite . as remarked in (
* section 2.1 ) , if @xmath177 is of general type , then the subgroup @xmath18 is self - commensurating in @xmath167 . if @xmath177 is the i - bundle type and @xmath167 acts trivially on @xmath174 then @xmath178 .
so @xmath18 is self - commensurating in @xmath167 in both these cases . as commensurable subgroups have the same commensurators and @xmath168 we conclude in both these cases that @xmath8 is a finite index subgroup of @xmath18 as required .
it remains to consider the situation when @xmath177 is of twisted i - bundle type i.e. @xmath167 acts non - trivially on @xmath174 .
here it is no longer evident that @xmath18 is self - commensurating in @xmath167 and the proof that @xmath8 is a subgroup of @xmath18 is slightly different . recall that @xmath174 is a 2-element set and @xmath167 acts nontrivially on @xmath174 so that the stabiliser of any point in @xmath174 is isomorphic to @xmath18 .
this implies that @xmath18 has index @xmath32 in @xmath167 and by ( * ? ? ?
* proposition viii.10.2 ) , @xmath167 is a @xmath63 group .
we argue that in this case @xmath167 is non - orientable , using the following observation .
* observation * if @xmath179 is an orientable @xmath83 pair of @xmath175-bundle type in which @xmath141 is an orientable @xmath63 group then @xmath141 acts trivially on @xmath174 .
the proof given here of the observation above is due to jonathan cornick and peter kropholler , and we are grateful to them for allowing us to include it .
consider the long exact sequence of cohomology for the pair with coefficients in the integral group ring @xmath180 .
cohomology groups with these coefficients inherit an action of @xmath141 .
we obtain the following short exact sequence @xmath181 the orientability of the pair @xmath179 implies @xmath182 is isomorphic to the trivial module @xmath183 .
the orientability of @xmath141 implies that @xmath184 is isomorphic to the trivial module as well . from this
it follows that the action of @xmath141 on the middle group @xmath185 is either trivial or of infinite order .
let @xmath8 be the stabiliser of one of the points of @xmath174 .
let @xmath37 be the unique maximal orientable @xmath63 subgroup in @xmath8 .
the action of @xmath37 on @xmath185 is trivial and @xmath37 has index at most @xmath186 in @xmath141 . from this
it follows that @xmath141 acts trivially on @xmath185 .
in particular ,
@xmath141 acts trivially on @xmath174 .
this establishes the observation . returning again to the proof of theorem [ pd ]
, we apply the observation with @xmath187 .
as the pair @xmath173 is of twisted @xmath175-bundle type , @xmath167 must be non - orientable , so it has a non - trivial dualising module @xmath188 .
proposition viii.10.2 of @xcite says that any @xmath63 subgroup of @xmath167 inherits its dualising module by restriction from the dualising module of @xmath167 , so @xmath167 has a unique maximal orientable @xmath63 subgroup , @xmath189 .
since @xmath167 is not orientable @xmath189 is a proper subgroup and since @xmath18 is orientable @xmath190 .
finally since the index @xmath191=2 $ ] we conclude that @xmath192 . since @xmath8 is also orientable , @xmath193 as required .
_ in this section ( and in particular in lemmas [ belowmid ] and [ middim ] below ) , for a smooth ( respectively , pl or topological ) manifold , _ submanifold _ means a smooth ( respectively , pl locally flat or topological locally flat ) submanifold .
_ let @xmath194 be manifolds as in the statement of theorem [ main ] and @xmath31 a @xmath2-injective map .
then @xmath195 satisfy the hypotheses of theorem [ dualitytheorem ] providing a splitting of @xmath28 as either an amalgamated free product @xmath196 or as an hnn extension @xmath197 where @xmath198 is a subgroup of finite index . in @xcite
cappell provided tools to geometrise such a splitting .
in particular he proved that if @xmath18 is square root closed in @xmath28 then there is a closed aspherical embedded submanifold @xmath199 such that @xmath64 induces the inclusion of @xmath18 in @xmath28 and , by van kampen s theorem , induces the splitting of @xmath28 .
asphericity then yields the homotopy commutative figure [ diagram2 ] of the introduction .
the idea of cappell s proof is to use surgery techniques to build a cobordism from @xmath26 to a manifold @xmath200 which does split in the required way .
this provides the homotopy equivalences @xmath29 required by theorem [ main ] and cappell s square root closed hypothesis on @xmath18 can be used to promote the homotopy equivalences to a homeomorphism @xmath201 , bypassing the need for @xmath28 to satisfy the borel conjecture .
since we prefer not to invoke either the borel conjecture or the square root closed hypothesis in our statement of theorem [ main ] it is necessary to unpack the proof of cappell s splitting theorem . as usual , surgery below the middle dimension requires no additional hypotheses , as is captured by the following lemma .
we have amended the notation to fit our situation : ( * ? ? ?
* lemma i.1 ) [ belowmid ] let @xmath202 be an @xmath203 dimensional poincar complex and @xmath36 a codimension-@xmath30 sub - poincar complex with trivial normal bundle in @xmath202 and with @xmath204 injective .
let @xmath26 be an @xmath203 dimensional closed manifold with @xmath205 a homotopy equivalence , @xmath206 .
assume we are given @xmath207 ; then @xmath46 is homotopic to a map , which we continue to call @xmath46 , which is transverse regular to @xmath36 ( whence @xmath208 is a codimension-1 submanifold of @xmath26 ) and such that the restriction @xmath209 induces isomorphisms @xmath210 , @xmath211 .
lemma [ middim ] describes the obstruction to carrying out surgery in the middle dimension which by lemma [ aspherical ] is all that is then required .
it is carried by the surgery kernels @xmath212 defined in ( * ? ? ? * section i.4 ) , and the projective class group @xmath17 appearing in lemma [ middim ] below . as recorded in the proof of (
* lemma i.2 ) , if @xmath213 is an isomorphism then @xmath214 . [ middim ]
let @xmath215 , @xmath26 be a closed manifold and @xmath202 a poincar complex of dimension @xmath203 .
assume we are given @xmath36 a sub - poincar complex ) of dimension @xmath14 of @xmath202 with trivial normal bundle and with @xmath216 .
assume further that @xmath205 is a homotopy equivalence transverse regular to @xmath36 with , writing @xmath208 , @xmath217 connected and @xmath218 an isomorphism and @xmath214 , @xmath219 . then letting @xmath220
denote the decomposition of @xmath221 modules defined in @xcite i.4 , 1 .
@xmath222 is a stably free @xmath221 module and @xmath223=-[q]$ ] .
moreover , + in case i , @xmath223\in \ker{(\widetilde{k_0}(c ) \rightarrow \tilde{k}_0(g_1 ) \oplus \tilde{k}_0(g_2))}$ ] , and + in case ii , @xmath223 \in \ker{(\widetilde{k_0}(c ) \rightarrow \widetilde{k_0}(j))}$ ] .
any finite set of elements of @xmath153 ( respectively , @xmath40 ) can be represented by embedded disjoint framed spheres in @xmath217 for @xmath224 .
the intersection pairing of @xmath222 is trivial when restricted to @xmath153 ( resp ; @xmath40 ) and @xmath225 .
thus , @xmath223=-[p^*]$ ] .
[ vanishingp ] if @xmath223=0 $ ] , @xmath46 is homotopic to a map @xmath226 with @xmath227 @xmath228 @xmath5-connected and so that , abusing notation by writing @xmath229 for the decomposition of @xcite i.4 , @xmath153 and @xmath40 are free @xmath221-modules .
applying lemma [ belowmid ] up to dimension @xmath230 ensures the vanishing of the surgery kernels @xmath212 while hypothesis [ k_0 ] of theorem [ main ] ensures the vanishing of @xmath17 so that @xmath223=0 $ ] as required by condition [ vanishingp ] of lemma [ middim ] .
cappell s lemmas along with his nilpotent normal cobordism construction allow us to realise the splitting obtained in theorem [ pd ] by an embedded submanifold @xmath217 in a homotopic manifold @xmath200 which is aspherical up to the middle dimension .
given that we are working with poincar duality groups the following elementary lemma from homotopy theory guarantees that this is sufficient to ensure that @xmath217 is aspherical .
[ aspherical ] let @xmath231 be a closed , orientable manifold and let @xmath5 be the largest integer less than or equal to @xmath232 .
suppose the universal cover @xmath233 of @xmath36 is @xmath5-connected and that @xmath234 is a duality group of dimension @xmath14 .
then @xmath235 for all @xmath236 .
let @xmath36 be a manifold as in the statement of the lemma .
suppose that @xmath237 for some @xmath236 and choose the smallest such @xmath238 .
evidently , @xmath233 is @xmath239-connected . by the hurewicz theorem ,
@xmath240 for all @xmath241 and @xmath242 @xmath243 .
observe that @xmath244 .
duality for orientable non - compact manifolds implies that @xmath243 @xmath245 . here , @xmath246 refers to the cohomology with compact supports for @xmath233 .
we claim that @xmath247 .
recall that @xmath248 @xmath249 , as @xmath37 varies over compact subsets of @xmath233 and hence is the @xmath64-th homology of the complex @xmath250 @xmath251 . note that every element of @xmath250 @xmath251 is represented by a module homomorphism from @xmath252 to @xmath86 for some compact @xmath37 . as @xmath7 acts freely and properly on the universal cover @xmath233 with quotient @xmath36
, we can consider the partial augmented singular chain complex @xmath253 .
since @xmath233 is @xmath239-connected this is a partial free @xmath7-resolution of @xmath86 and may be completed to a projective resolution @xmath153 .
the action of @xmath7 is proper and so for @xmath254 , the cohomology of the co - complex @xmath255 is the cohomology of @xmath36 with local coefficients in @xmath256 .
hence this is the @xmath64-th cohomology of @xmath7 with @xmath256 coefficients .
we now show that for @xmath257 , @xmath258 @xmath259 , where @xmath37 varies over compact subsets of @xmath233 .
let @xmath260 . as a @xmath86-module map , @xmath261 @xmath262
is induced by the assignment @xmath263 , where the sum is finite and @xmath264 is a module homomorphism from @xmath265 to @xmath86 .
a straightforward computation shows that @xmath266 is a @xmath7-module homomorphism if and only if @xmath267 for all @xmath268 .
note that @xmath269 is zero outside of a compact subset @xmath37 of @xmath233 and hence we have a map @xmath258 @xmath270 given by @xmath271 .
one checks that this is an isomorphism with inverse coming from the prescription @xmath272 and hence @xmath273 . finally , putting all the isomorphisms together , @xmath274 . as
@xmath275 and @xmath7 is a duality group of dimension @xmath14 group , @xmath276 except possibly when @xmath277 , so in particular @xmath278 which contradicts our hypothesis on @xmath279 .
let @xmath26 and @xmath6 be as in the statement of theorem [ main ] . set @xmath280 and @xmath35 .
then @xmath7 is an orientable @xmath281 group and @xmath8 is an orientable @xmath83 subgroup of @xmath7 and by theorem [ pd ] , @xmath7 splits over a poincar duality subgroup @xmath18 containing @xmath8 as a subgroup of finite index .
it is well known that a group @xmath7 splits over a subgroup @xmath8 ( in the sense of our convention ) if and only if @xmath7 acts on an unbounded tree @xmath49 with a single edge orbit , and so that @xmath8 is an edge stabiliser .
it follows from the unboundedness of the action that vertex and edge stabilisers of the action are all of infinite index in @xmath7 . when the splitting is a non - trivial amalgamated free product @xmath282 , the subgroups @xmath283 and @xmath18 are all of infinite index in @xmath7 and
so , by strebel s theorem ( * ? ? ?
* theorem 3.2 ) , the cohomological dimension of each of @xmath284 is at most @xmath14 . since @xmath166 contain @xmath18 which has cohomological dimension @xmath14 , all three have cohomological dimension equal to @xmath14 .
similarly , in the case of the hnn decomposition , the subgroups @xmath285 have cohomological dimension @xmath14 . since @xmath286 , the eilenberg - ganea theorem @xcite , applies and the cohomological dimensions of @xmath167 , @xmath125 and @xmath18 are equal to the geometric dimensions of @xmath167 , @xmath125 and @xmath18 respectively .
we now carry out the standard mapping cylinder construction of an eilenberg maclane space for @xmath7 .
when the group @xmath7 splits as an amalgamated free product @xmath287 let @xmath288 , @xmath289 and @xmath290 denote the @xmath14-dimensional @xmath291 complexes for @xmath167 , @xmath125 and @xmath18 .
note that these complexes may be taken to be simplicial complexes .
since @xmath292 are both of type fp , we can apply the mayer vietoris sequence , to see that the groups @xmath167 and @xmath125 are also of type fp so we may choose the complexes to be of finite type .
let @xmath293 and @xmath294 be maps inducing the inclusions @xmath295 and @xmath296 .
we construct the mapping cylinders @xmath297 of these maps and making the standard identifications yields a @xmath291 for the group @xmath7 which we will denote by @xmath202 . in the case
when @xmath7 splits as an hnn extension we carry out a similar process building a @xmath298 from @xmath288 and @xmath299 $ ] by glueing the two ends @xmath300 to @xmath288 using maps which induce the two inclusions of @xmath18 into @xmath167 defining the hnn extension . in this case
again we denote the @xmath291 by @xmath202 .
note that @xmath202 is a poincar complex , and moreover , the poincar subcomplex @xmath301 cuts the neighbourhood @xmath302 into two components so that the normal bundle of @xmath290 in @xmath202 is trivial ( see introduction of @xcite ) . by construction
, the composition @xmath303 is @xmath2-injective .
applying lemma [ belowmid ] , we may replace @xmath46 ( up to homotopy ) by a map , which we continue to call @xmath46 , which is transverse regular to @xmath304 and such that the restriction of @xmath46 to @xmath305 induces isomorphisms @xmath306 , for all @xmath307 .
in particular , @xmath308 for all @xmath309 . note that the transverse regularity of @xmath46 ensures that the pre - image @xmath217 is a codimension-@xmath30 submanifold of @xmath26 .
our aim is to further modify the map @xmath46 so that it remains transverse regular and so that its restriction to @xmath217 induces an isomorphism on all homotopy groups , thus making @xmath217 aspherical .
recall that @xmath14 is even so that @xmath215 with @xmath310 . as recorded in the proof of lemma i.2 of @xcite , the isomorphisms @xmath306 ,
for all @xmath311 ensure that each of the modules @xmath212 for @xmath312 vanishes .
since @xmath18 is a torsion free finite extension of @xmath8 , condition [ k_0 ] of theorem [ main ] ensures the vanishing of @xmath17 . invoking lemma [ middim ]
, we replace @xmath46 by a map @xmath226 homotopy equivalent to @xmath46 which is transverse regular and such that @xmath313 where @xmath314 and the restriction of @xmath226 to @xmath217 is @xmath5-connected .
in addition , both @xmath153 and @xmath40 in @xmath229 are free @xmath18-modules .
this means that one can perform cappell s nilpotent normal cobordism construction on @xmath26 , a process we will now describe . note
that cutting @xmath26 along @xmath217 we get a decomposition @xmath315 in case i and @xmath316 in case ii .
the covering of @xmath26 corresponding to the image of @xmath317 is labelled @xmath318 and the universal covering of @xmath26 is labelled @xmath47 .
the group @xmath319 acts by covering transformations on @xmath45 , @xmath320 and @xmath321 ( where @xmath322 ) with quotients being @xmath318 , @xmath323 and @xmath324 .
hence , @xmath325 . let @xmath326 $ ] and @xmath327 $ ] .
we select a tubular neighbourhood @xmath328 of @xmath217 in @xmath26 such that when the lift of @xmath217 to @xmath318 is extended to a lift of @xmath328 , we have @xmath329 and @xmath330 . via surgery along the spheres @xmath331 and @xmath332.,width=576 ] let @xmath333 denote a @xmath334-free basis for @xmath153 and let @xmath335 denote the dual basis for @xmath40 under the intersection pairing of @xmath222 .
we choose disjoint framed spheres @xmath336 and @xmath337 representing the bases for @xmath153 and @xmath40 respectively .
we can assume given the choice of bases , that for @xmath338 , @xmath339 and for any @xmath64 , @xmath340 intersects @xmath341 in a point .
we kill the spheres @xmath340 by @xmath5-surgery on @xmath217 to obtain a new manifold @xmath342 yielding a cobordism @xmath343 of @xmath217 with @xmath342 ; similarly killing the spheres @xmath341 by @xmath5-surgery we get a cobordism @xmath344 of @xmath217 with a new manifold @xmath345 .
clearly , @xmath346 and by lemma [ aspherical ] both @xmath347 are aspherical and thus homotopy equivalent to @xmath290 .
now consider the trivial cobordism @xmath348 .
we will extend it to a cobordism from @xmath26 to @xmath200 by applying the cobordism extension lemma , gluing @xmath349 $ ] to @xmath350 along @xmath351 \times \{1\}$ ] and glueing @xmath352 $ ] to @xmath350 along @xmath353 $ ] .
the boundary of the resulting manifold @xmath49 has two components , one being @xmath26 and the other , a new manifold @xmath200 given as follows .
+ @xmath354 + @xmath355 ) \cup_{n'_p } t_p \cup_{n ' } ( n ' \times [ -1,1 ] ) \cup_{n ' } ( t_q \cup_{n'_q } n'_q \times [ 1,2]).$ ] observe that the manifold @xmath356 sits naturally inside @xmath200 and gives a cobordism between @xmath342 and @xmath345 .
moreover not only are @xmath342 and @xmath345 homotopy equivalent to @xmath290 , the cobordism @xmath356 constructed by filling the spheres @xmath357 and @xmath358 is also homotopy equivalent to @xmath290 .
the original homotopy equivalence @xmath359 extends to a map @xmath360\cup
t_q\times [ 1,2]\rightarrow y\times i$ ] which restricts first to a homotopy equivalence @xmath361 ( ( * ? ? ?
* lemma ii.5 ) ) and further to the homotopy equivalence @xmath362 .
hence we obtain a manifold @xmath200 which is homotopy equivalent to @xmath26 and split along @xmath363 via the embedded submanifold we continue to call @xmath217 .
recall also that @xmath8 is a finite index subgroup of @xmath18 .
hence , there is a finite cover @xmath364 of @xmath217 with fundamental group @xmath8 .
as @xmath364 and @xmath6 are aspherical manifolds with isomorphic fundamental groups there exists a homotopy equivalence @xmath365 between them .
this completes the proof of theorem [ main ] .
recall the following classical fact : if @xmath65 is a smooth manifold such that the first betti number @xmath67 of @xmath26 is @xmath366 and @xmath68 has non - zero signature then there are no immersions of @xmath6 into @xmath26 .
this follows using hirzebruch s signature theorem : since @xmath46 is a codimension-1 immersion , @xmath367 maps the hirzebruch l - class @xmath368 onto @xmath369 .
it follows from poincar duality that @xmath370 is isomorphic to the torsion free part of @xmath371 .
therefore , the vanishing of @xmath67 forces @xmath369 to be zero .
however hirzebruch s signature theorem says that @xmath369 is equal to the signature of @xmath6 .
this applies , for example , to show that there are no codimension-1 immersions of an orientable quaternionic hyperbolic or cayley hyperbolic manifold @xmath68 into a smooth orientable aspherical manifold @xmath65 with @xmath372 .
appealing instead to the non - vanishing of pontryagin numbers , we obtain the following generalisation as a corollary to theorem [ main ] , which obstructs the existence of @xmath2-injective continuous maps rather than immersions . note
that while our hypotheses on @xmath6 imply that it satisfies the borel conjecture so that we can take the homotopy equivalence @xmath365 to be a homeomorphism , we do not assume in the statement that the target manifold satisfies the borel conjecture , nor do we assume cappell s square root closed hypothesis , since in the proof we are applying theorem [ main ] in its most general form .
[ pontryagin ] let @xmath65 be a closed , orientable , aspherical , smooth manifold such that @xmath66 and the first betti number @xmath67 is zero .
let @xmath68 be a closed , orientable , aspherical , smooth manifold with at least one non - zero pontryagin number such that @xmath16 is either word hyperbolic or a cat(0 ) group and satisfies kazhdan s property ( t ) .
then there are no @xmath2-injective continuous maps @xmath69 . carrying out the cappell surgery arguments in the smooth category , the proof of theorem [ main ] shows that there is a 2-sided embedded smooth submanifold @xmath217 in a smooth manifold @xmath200 homotopy equivalent to @xmath26 that realises the splitting .
furthermore , since @xmath16 satisfies the borel conjecture , the map @xmath373 is a covering map . now
if @xmath217 is separating , write @xmath374 for one of the connected components of the manifold obtained from cutting @xmath200 along @xmath217
. then @xmath374 is a smooth orientable manifold with boundary @xmath217 and hence @xmath217 bounds orientably .
this means that all pontryagin numbers for @xmath217 must vanish . on the other hand
we are assuming that @xmath6 has at least one non - zero pontryagin number ( for the case when @xmath6 is quaternionic or cayley hyperbolic , see @xcite ) .
since pontryagin numbers vary multiplicatively with degree on covering maps the pontryagin numbers of @xmath217 are also non - zero and so @xmath217 can not bound orientably .
this means that @xmath217 can not be separating and so intersection with @xmath217 yields an element of infinite order in the first cohomology of @xmath200 . by duality and the universal coefficients theorem the first betti number @xmath375 and
hence @xmath67 is non - zero .
bartels , a. and lck , w. , the borel conjecture for hyperbolic and cat(0 ) groups .
ann . of math .
( 2 ) 175 ( 2012 ) , no .
2 , 631689 .
bieri , r. , homological dimension of discrete groups , queen mary college mathematics notes , london ( 1976 ) .
brown , k. , cohomology of groups .
graduate texts in mathematics 87 , springer - verlag , new york - berlin ( 1982 ) .
cappell , s. , a splitting theorem for manifolds , invent . math .
33 , 69170 ( 1976 ) .
davis , m. , the geometry and topology of coxeter groups , london mathematical society monographs , princeton university press ( 2008 ) .
dicks , w. and dunwoody , m. j. , groups acting on graphs , cup ( 1982 ) .
dixon , j. d. and mortimer , b. , permutation groups , graduate texts in mathematics 163 , springer ( 1996 ) .
dunwoody , m. j. and sageev , m. e. , jsj - splittings for finitely presented groups over slender groups .
inventiones mathematicae , 135 , ( 1 ) , 2544 ( 1999 ) .
( doi:10.1007/s002220050278 ) .
dunwoody , m. j. and swenson , e. l. , the algebraic torus theorem .
inventiones mathematicae , 140 , ( 3 ) , 605637 ( 2000 ) .
( doi:10.1007/s002220000063 ) .
eilenberg , s. and ganea , t. , on the lusternik - schnirelmann category of abstract groups , ann . of math .
65 , 517518 ( 1957 ) .
farb , b , the quasi - isometry classification of lattices in semi - simple lie groups , mathematical research letters 4 , 705717 ( 1997 ) .
farrell , f. t. and jones , l. e. , topological rigidity for compact non - positively curved manifolds , differential geometry : riemannian geometry ( los angeles , ca , 1990 ) , 229274 , amer .
soc . , providence , ri , ( 1993 ) .
kahn , j. and markovic , v. , immersing almost geodesic surfaces in a closed hyperbolic three manifold , ann . of math .
( 2 ) 175 ( 2012 ) , no .
3 , 11271190 . kropholler , p. h , a group theoretic proof of the torus theorem , geometric group theory volume 1 , eds .
niblo and m. roller , lms lecture notes series 181 , cup ( 1993 ) 138158 .
kropholler , p.
h. and roller , m. a. , splittings of poincar duality groups iii , j. london math .
39 , 271284 ( 1989 ) .
lafont , j. and roy , r. , a note on the characteristic classes of non - positively curved manifolds , expo .
25 , 2135 ( 2007 ) .
mok , n. , siu , y. t. and yeung , s. , geometric superrigidity .
113 , 5783 ( 1983 ) .
niblo , g. a , the singularity obstruction for group splittings , topology appl . 119 , 1731 ( 2002 ) .
niblo , g. a. and reeves , l. d. , groups acting on cat(0 ) cube complexes , geom .
topol . 1 , 18 ( 1997 ) . papakyriakopoulos , c. , on dehn s lemma and asphericity of knots .
ann . of mathematics 66 ( 1 ) , 126 ( 1957 ) .
doi:10.2307/1970113 . jstor 1970113 reid , a. , a non - haken hyperbolic 3-manifold finitely covered by a surface bundle , pacific j. math .
167 , 163182 ( 1995 ) .
sageev , m. , co - dimension 1 subgroups and splittings of groups , j. algebra , 189 , 377389 ( 1997 ) .
scott , g. p. , ends of pairs of groups , journal of pure and applied algebra , 11 ( 1977 ) 179198 .
scott , g. p. and swarup , g. a. , regular neighbourhoods and canonical decompositions for groups .
asrisque , ( 289):vi+233 , ( 2003 ) .
strebel , r. , a remark on subgroups of infinite index in poincar duality groups , comment .
52 , 317324 ( 1977 ) .
waldhausen , f. , on the determination of some bounded 3-manifolds by their fundamental groups alone , proc .
sympos . on topology and its applications , herieg - novi , yugoslavia , aug . 25 - 31 , 1968 , belograd , 1969 , 331332 . | we show that for a wide class of manifold pairs @xmath0 satisfying @xmath1 , every @xmath2-injective map @xmath3 factorises up to homotopy as a finite cover of an embedding .
this result , in the spirit of waldhausen s torus theorem , is derived using cappell s surgery methods from a new algebraic splitting theorem for poincar duality groups . as an application we derive a new obstruction to the existence of @xmath2-injective maps . |
in the following , we consider the optimization of a state - to - state transfer problem where the figure of merit is the overlap of the final state after the evolution with the target state @xmath2 here @xmath3 is the target state and the time evolution is given by the schrdinger equation @xmath4 and control function @xmath5 is determined by the optimization algorithm in order to maximize the figure of merit .
the crab algorithm @xcite builds on the fact that in most scenarios the resources available to solve an optimal control problem such as time , energy and bandwidth are limited : in particular , as the set of practically accessible wave functions is usually limited , one can show that also the control bandwidth of the control field can be upper bounded @xcite .
this bound has very important consequences as optimal control problems can be practically solved by exploring a small subset of the a priori infinite dimensional search space of functions .
that is , one can expand the control field in a truncated basis @xmath6 and the optimization can then be performed on this subspace of _ small _ dimension , resulting in an optimal set of coefficients @xmath7 .
the optimization can be performed by standard tools , e.g. by the nelder - mead simplex algorithm that does not rely on gradients @xcite . a standard choice for the basis functions @xmath8 are trigonometric functions , often multiplied by a shape function @xmath9 that fixes the pulse boundary conditions and a guess pulse @xcite . due to the restriction of the search basis to @xmath10 dimensions given by the crab expansion in eq .
, the algorithm might converge to a non - optimal fixed point , i.e. the algorithm is trapped in a local minimum arising due to the constraint - a so - called false trap . to overcome this problem , and escape from these false traps , we show in the next section that one can start from the non - optimal fixed point a new crab optimization with a new random basis and new coefficients .
this is done in an iterative way so that in the @xmath11-th super - iteration one optimizes the coefficients @xmath12 of @xmath13 where @xmath14 are new randomly chosen basis functions , for example sine or cosine functions with random frequencies within some interval @xmath15 $ ] . as a consequence , in each super - iteration the old pulse is dressed with new search directions and we call this procedure dressed chopped random basis ( dcrab ) algorithm @xcite .
this updating of the search directions can also be understood as an extension of powell s method @xcite to an infinite dimensional search space . in the following sections we give a theoretical explanation why this is a substantial improvement of the algorithm by analyzing how it influences the constrained control landscape and we demonstrate it numerically by applying it to a model with typical properties . .
while @xmath16 has a strict global maximum point at @xmath3 ( left panel ) , this state can be reached by different control functions @xmath17 and @xmath18 corresponding to multiple global maximum points in the landscape @xmath19 ( right panel).,scaledwidth=50.0% ]
in this and in the following section , we review the theory of control landscapes and how it can enlighten the reasons why optimal control algorithms converge to the optimal solution or become trapped @xcite .
we first review a general perturbative analyses of control landscapes presented in @xcite , specifically its gradients and critical points , valid independently from the particular optimization algorithm employed to find the optimal driving field ; and then we specify it to the crab approach we are focusing here . in particular , we focus on why crab can be trapped whereas dcrab like gradient methods can not be trapped , as we will show also in the next section by means of some numerical examples . given a control problem with control @xmath5 , the control landscape is the functional @xmath20 with @xmath21 where @xmath22 is the final state resulting from time evolution of the initial state @xmath23 with the given control @xmath5 and @xmath16 is a figure of merit quantifying the quality of the process @xcite . in other words ,
the control landscape is the functional expressing the figure of merit of the process as a function of the control field . in the following we use the state fidelity given by eq . as a figure of merit .
notice that the control landscape can be seen either as a function of the final state @xmath24 or as a functional of the control field @xmath5 : in the first case the function has by definition a single optimal point ( @xmath25 , up to a global phase ) , while in the latter there might exist different control pulses leading to the same maximum value , as sketched in fig .
[ fig : landscape ] .
optimization usually leads to so - called critical points of the landscape , that is to pulse shapes fulfilling the condition @xmath26 i.e. a vanishing variation of the functional @xmath19 for a variation of the control @xmath27 . using the chain rule one can show that this variation consists of two parts : the gradient of the fidelity as a function of the final state @xmath28 , and the variation of the final state as a result of the variation of the control @xmath29 .
for all common choices of the figure of merit a vanishing gradient @xmath30 corresponds to the global maximum , global minimum or a saddle point @xcite ; more specifically if @xmath31 is the state overlap ( eq . )
there are no saddle points since @xmath32 corresponds to @xmath33 ( the global minimum ) or @xmath34 ( the global maximum ) .
less intuitive is the role played by the second term , which we analyze in the following , to understand under which conditions @xmath35 implies @xmath30 .
critical points can be classified as singular or regular critical points , where regular means that for every @xmath36 in the hilbert space of the problem there is a change in the control @xmath37 that generates it , while on the contrary for singular points this is not the case @xcite . as a direct consequence , for regular critical points we have @xmath38 as in this case
the gradient of the fidelity is orthogonal to the whole hilbert space .
this means that a regular point is not a trap .
that is , all traps have to be singular critical points ( although the reverse is not necessarily true ) .
however , at least for controllable systems , all known traps occur at constant control fields @xcite , often at @xmath39 , and numerical evidence hints to the fact that the singular points at non - constant control seem to be no traps @xcite . that is , it is commonly accepted that unconstrained control landscapes for all practical purposes have no traps , although a rigorous proof has been given only for two specific situations , the landau - zener system @xcite and the transmission of a wave package over a potential barrier @xcite .
now we discuss the implications of these results on the crab and dcrab optimal control success rate . as a starting point , we recall that the previous statements assume that the control @xmath27 is a general function in @xmath40 whereas for crab we are much more restricted by the expansion into the truncated basis .
this means that one might find false traps in the crab control landscape , i.e. encounter points where @xmath41 vanishes for all variations @xmath37 allowed by the truncated basis expansion of eq . , but
not for all variations of the unconstrained control space : @xmath42 these points are called false traps @xcite as they arise only artificially from the choice of the basis and their influence can hinder convergence of the algorithm @xcite . here ,
we show how these false traps for crab are removed by the super - iterations of dcrab : from equations ( [ eq : criticalpoint - chainrule ] ) and ( [ eq : pseudocriticalpoint ] ) together with @xmath43 it follows that , for a false trap the gradient has to be @xmath44 for some non - zero vector @xmath45 .
a perturbative treatment yields @xcite : @xmath46 this results in the expression for the overlap of the state update and the gradient @xmath47 where @xmath48 is a continuous function and @xmath49 since @xmath50 would violate eq .
( [ eq : pseudocriticalpoint ] ) .
if we now choose @xmath51 with a new random frequency @xmath52 we get @xmath53 almost surely ( i.e. the integral vanishes only on a null set of the probability measure ) .
if we now perturb our control field @xmath27 by this new frequency contribution ( this is the new super - iteration of dcrab ) we find @xmath54 and thus we have removed the false trap . in conclusion ,
the recipe to escape from a false trap is simply to add a new random frequency term to the crab expansion once we are in the false trap .
this can be done also without increasing the total number of coefficients since when one is at the bottom of a false trap for a given set of basis functions and coefficients there is no more use in varying the old coefficients , and they can be kept at their value .
we thus use the pulse @xmath27 leading to the false trap as a guess pulse for the next super - iteration of dcrab ( as explained in the previous section , see eq .
( [ eq : dcrab ] ) ) that opens the false trap .
in this section we show that by choosing enough new frequencies in a single super - iteration of dcrab we can follow approximately the instantaneous gradient of the control landscape with a bandwidth - limited pulse update . in the previous section we chose a new basis function so that its scalar product with @xmath55 would be finite .
it is worth noting that this function @xmath56 is exactly the update direction used in gradient algorithms with the time - evolved initial state @xmath57 and the adjoint state @xmath58 together with @xmath59 we get the well - known @xmath60 to follow the gradient of the control landscape only a limited freedom for the pulse update is needed .
hsieh et al .
@xcite derived for a state to state transfer in an @xmath0-level system a set of @xmath61 functions ( for each iteration step ) that span the pulse update @xmath56 .
this simply reflects the fact that the state update generated by the pulse update @xmath37 , i.e. @xmath36 , is a @xmath0 dimensional complex vector ( @xmath62 real coefficients ) with norm @xmath63 and irrelevant global phase ( thus @xmath61 real coefficients ) .
we can generate something very similar by choosing @xmath64 random frequencies ( we do not fix the global phase here ) and corresponding pulse updates @xmath65 ( @xmath66 ) with real coefficients @xmath67 .
let us here assume that @xmath40 pulses can generate @xmath36 out of the whole tangential space of the unity sphere of @xmath68 ( i.e. regularity of the control landscape at the point of consideration which is the general case for trap free landscapes ) .
the state update generated by each pulse update in a perturbative regime reads @xcite : @xmath69 we now show that for every vector @xmath70 out of this unit sphere there is a set of real coefficients @xmath71 so that @xmath72 . in order to do
so we identify @xmath73 and @xmath74 as elements of a @xmath64 dimensional real vector space with scalar product @xmath75 and show that the @xmath74 are a basis of this space .
let us thus consider @xmath76 .
we will show by induction that @xmath77 has dimension @xmath78 . for @xmath79
this is trivial .
if now @xmath77 has dimension @xmath78 we have to show that @xmath80 is not orthogonal to @xmath81 .
let thus be @xmath82 .
the overlap of @xmath80 and @xmath83 is @xmath84 the kernel @xmath85 is continuous and if @xmath86 the overlap is nonzero almost surely .
however , @xmath87 @xmath88 @xmath89 ( @xmath90 ) @xcite and thus @xmath83 is not in the tangential space .
this proves that the @xmath74 as span the whole @xmath64 dimensional real vector space , or in other words the @xmath74 with real coefficients span the whole tangential space of the unit sphere of @xmath68 ( almost surely ) .
we can also orthogonalize the state updates in a gram schmidt way to obtain the orthogonal basis @xmath91 .
this translates into a change of the pulse updates via @xmath92 with @xmath93 generated by @xmath94 .
each direction @xmath95 is then a linear superposition of the basis states @xmath96 and can thus be generated by a superposition of the corresponding pulses @xmath97 in particular this proves that with @xmath64 random instantaneous basis functions ( @xmath98 oder @xmath94 for the `` orthogonal '' basis ) we can follow the instantaneous gradient of the control landscape by suitable coefficients . we stress that , unlike in the case of a standard gradient method , this is done by bandwidth - limited pulses which can be comfortably adapted to typical experimental constraints such as bandwidth - limited control electronics .
this result demonstrates that after a change of the truncated basis in the function space the search in the state space is quasi - local .
we can use this in the numerical part by initializing a small simplex after the basis change to ensure local search .
the orthogonalization procedure could also help to find a better basis for the crab ( re)start at the cost of @xmath99 additional function evaluations .
in this section we show that the dcrab algorithm is indeed capable of escaping from false traps in random optimal control problems of increasing complexity .
we consider a spin hamiltonian @xmath100 with random coefficients @xmath101 , @xmath102 $ ] and a control field @xmath5 tuning the interaction .
the random coefficients lift the symmetry of the system and make it controllable @xcite .
we investigate the control resources needed to drive the system from a random initial state @xmath103 to a random target state @xmath3 within a fixed time interval @xmath104 $ ] .
we measure the fidelity of this state - to - state transfer by eq . .
we investigate different system sizes and different operation time intervals , analyze the occurrence of false traps in the landscape by several trials of optimization of the fidelity @xmath16 for different random instances of @xmath105 and @xmath106 ( uniformly distributed over the unit sphere ) , and measure the success of an optimization by a certain threshold @xmath107 for the residual error @xmath108 .
we count an optimization trial as a success if @xmath109 after convergence of the optimization or after a certain maximum number of function evaluations .
the success rate of the optimization depends on the presence or absence of false traps in the landscape . to detect them
we optimize ten different sets of random instances of the hamiltionian and of the initial and goal state ; for each instance we use ten different random starting points and random sets of frequencies to perform the crab optimization .
the frequencies of the basis functions ( both for crab and dcrab ) are chosen randomly out of an interval @xmath110 $ ] : we choose @xmath111 for @xmath112 respectively . as a function of the number @xmath10 of basis functions in the crab expansion . as we increase the size of the function space
false traps are removed .
the symbols are the mean values over 10 different random pairs of final and initial states @xmath103 and @xmath3 with 10 different starting points and frequencies each .
the error bars show the standard deviation over the different final and initial states .
the total time was set to @xmath113 and the allowed bandwidth to @xmath111 for @xmath112 qubits respectively .
an optimization was counted as success when the residual error @xmath114 was smaller than @xmath107 .
the black lines indicate the empirical `` @xmath115 '' rule for the required number of coefficients @xmath10 .
, scaledwidth=47.0% ] fig .
[ fig : nt - vs - freq ] reports the success probability @xmath116 for the standard crab optimization as a function of the number @xmath10 of coefficients of the truncated basis expansion in eq . : as can be clearly seen , for @xmath10 large enough no false traps are present , resulting in a success probability of one . here , by large enough we mean the empirical `` @xmath117 '' rule for unconstrained optimization , where the number of real coefficients equals the number of independent real entries in the state vector @xcite .
the correspondent analysis for the dcrab approach is reported in fig .
[ fig : traps - removed ] : the success probability @xmath118 is always one regardless of @xmath10 , false traps are avoided and the fidelity always exceeds the threshold . despite the striking difference in terms of success probability , an important benchmark for any optimization method is given by the computational effort required to arrive at the optimal solution .
here we focus on the number of function evaluation needed to achieve the global optimum as this is practically the only difference between the two methods : fig .
[ fig : acrab-4qbt ] shows the number of function evaluations @xmath119 required by the two methods to exceed the threshold @xmath120 as a function of the coefficients @xmath10 in the crab expansion ( in the case of dcrab the coefficients of a single super - iteration ) .
all points consist of the average number of function evaluations of the successful runs divided by the respective success probability , that is the number of function evaluations that on average one has to do to solve the optimal control problem using one of the two methods .
note that the minimal effort does not follow the `` @xmath115 '' rule of fig .
[ fig : nt - vs - freq ] for guaranteed convergence . for crab
the computational effort heavily depends on @xmath10 , which can be problematic as the best choice of @xmath10 is not known in advance . for dcrab instead there is only a minor dependance on @xmath10 in the order of magnitude of the error bars . furthermore , fig .
[ fig : acrab-4qbt ] shows that even with the best choice of @xmath10 crab can not beat the performance of dcrab . as a function of the number @xmath10 of basis functions in a single call of crab within the dcrab super - iterations . independently of @xmath10 no false traps are encountered and @xmath121 .
the total time , allowed bandwidth and error threshold are as in fig .
[ fig : nt - vs - freq ] . , scaledwidth=47.0% ] for dcrab ( red ) and original crab ( black ) as a function of the number @xmath10 of basis functions involved in a single call of the respective algorithm .
the error bars show the logarithmic standard deviation .
optimization was stopped when the error @xmath114 crossed the threshold @xmath107 .
the total time was @xmath122 , the allowed bandwidth @xmath123 and the system size @xmath124 qubits .
similar results are obtained for other choices of the parameters .
, scaledwidth=47.0% ] finally , we study how the convergence properties change in the presence of additional constraints like limited fluence and bandwidth typically present in experimental setups that violate the hypothesis of the analysis of control landscapes presented up to now . indeed , in this scenario
, false traps might be present which might change performance and convergence speed of the algorithms .
we consider separately two different kinds of constraints : bandwidth - limited control and limited pulse height .
in the first case , we observe that eq . still holds even if the new random frequency @xmath52 is chosen only within the limited bandwidth interval @xmath15 $ ] .
we can then study the performance of the optimization as a function of @xmath125 , as done in the unconstrained case and compare the crab and dcrab approaches . as a function of the limited bandwidth @xmath126 for @xmath127 and @xmath128 in the system of @xmath124 qubits and total time @xmath122 with fixed random coefficients @xmath101
, @xmath129 in the hamiltonian ( eq . ) .
the grey diamonds show the infidelity @xmath114 for crab ( @xmath130 ) , while the green circles show it for dcrab ( @xmath131 ) .
the black dashed line indicates the error threshold of @xmath107 .
the error bars report the logarithmic standard deviation.,scaledwidth=47.0% ] the results are reported in fig .
[ fig : bandwidth ] where we show the optimal infidelity @xmath132 reached from ten independent runs as a function of @xmath126 .
the optimal control problem is to perform the state transfer from @xmath127 to @xmath128 given the hamiltonian of eq .
for crab ( grey diamonds ) and dcrab ( green dots ) .
one can see that for @xmath133 dcrab succeeds with probability one , as all instances reached the optimal fidelity .
notice that this is less than half the bandwidth of the previous results reported in figs .
[ fig : traps - removed ] and [ fig : acrab-4qbt ] .
in addition , in a small intermediate regime around @xmath134 , some optimizations succeed while others fail indicating the presence of false traps ( note that the graph shows just the mean value of the infidelity @xmath114 and the standard deviation ) ; while for @xmath135 the final state can not be reached anymore , indicating that the bandwidth is too small to achieve the desired result . in the case of crab ,
the three regimes are shifted toward larger frequencies .
the lower bound observed in both cases is in agreement with an information theoretical argument given in @xcite : to achieve full control over the system the inequality @xmath136 has to be fulfilled , where @xmath137 is the dimension of the state space .
this inequality basically says that the control has to contain enough information to distinguish the target state from the other reachable states and it yields @xmath138 , the value of the bandwidth where the infidelity in fig .
[ fig : bandwidth ] starts to drop indicating that control over the system starts to be effective .
we then perform a similar analysis for pulse height limited control with dcrab , where we study two scenarios to include such a constraint .
we first introduce a smooth constraint as usually done , that is a penalty on the pulse height so that the control objective becomes @xmath139 as a second alternative , we limit the pulse height by a hard wall constraint , by using the update formula @xmath140 the results of these two procedures are reported in fig . [
fig : pulseheight ] where we plot the infidelity @xmath132 as a function of the maximal pulse height @xmath141 : clearly the optimization works better with the hard boundaries than with the lagrange multiplier , as the fidelity threshold @xmath142 can be exceeded for about three times weaker pulses .
this difference can be understood by the fact that the hard wall introduces pulses of higher bandwidth .
compared to the unconstrained system we can decrease the maximal value of the pulse by a factor of fifteen , while keeping 100% success probability . for even smaller
cut - off @xmath143 the small error bars indicate that optimization failure is most probably more due to a loss of controlability than due to false traps . as a function of the limited pulse height @xmath141 for the transfer from @xmath144 to @xmath145 in the 2-qubit system with constrained dcrab .
the black line indicates the error threshold of @xmath107 .
the orange triangles are obtained with lagrange multipliers ( see eq . ) , while the brown circles are obtained with a cut - off at @xmath141 ( see eq . ) .
the error bars indicate the standard deviation over 10 different starting points of the optimization .
the orange triangles have also errorbars in @xmath143 since the lagrange multiplier is not a hard wall and instead we plot the maximum absolute value the optimal pulse takes in each realization . , scaledwidth=47.0% ]
we have generalized the results presented in @xcite to the case of bandwidth - limited control pulses , under the condition that the bandwidth satisfy the theoretical bound introduced in @xcite .
thanks to this theoretical result , we have modified the crab optimization algorithm to efficiently combine the advantages of gradient methods with those of truncated basis methods .
we showed that it is possible to exploit both the guaranteed convergence to the global optimum that gradient algorithms exhibit in the frequent case in which the kinematic ( @xmath31 ) and the dynamical ( @xmath20 ) landscapes are equivalent and the numerical gradient - free truncated basis approach .
moreover , we showed for two typical constraints , namely the limited fluence and the limited bandwidth constraints below the theoretical bound , that some of the convergence properties survive , if the constraints are carefully implemented in the optimization procedure .
we expect that the presented results will allow to tackle in the near future both theoretically and experimentally even more complex many - body problems as done so far , as well as a broader variety of control objectives and constraints .
10 f. schfer , i. herrera , s. cherukattil , c. lovecchio , f.s .
cataliotti , f. caruso , and a. smerzi , nat .
comm . 5 , 3194 ( 2014 ) . c. lovecchio , f. schfer , s. cherukattil , a. khan martuza , i. herrera , f.s .
cataliotti , t. calarco , s. montangero , and f. caruso , arxiv:1405.6918 .
i. bloch , nature 453 , 101622 ( 2008 ) .
p. schau , m. cheneau , m. endres , t. fukuhara , s. hild , a. omran , t. pohl , c. gross , s. kuhr , and i. bloch , nature 491 , 97 - 91 ( 2012 ) .
s. hoyer , f. caruso , s. montangero , m. sarovar , t. calarco , m.b .
plenio , k.b .
whaley , new j. phys .
16 045007 ( 2014 )
. f. caruso , s. montangero , t. calarco , s. f. huelga , and m. b. plenio , phys .
a 85 , 042331 ( 2012 ) .
y. o. dudin and a. kuzmich , science 336 , 887 ( 2012 ) .
h. kbler , j. p. shaffer , t. baluktsian , r. lw , and t. pfau , nature photon . 4 , 112 ( 2010 ) .
obrien , science 318 , 156770 ( 2007 ) .
p. kok , w.j .
munro , k. nemoto , t.c .
ralph , j.p .
dowling , and g.j .
milburn , rev .
79 , 135 ( 2007 ) .
s. schmidt , j. koch .
annalen der physik , 525 , 395 - 412 ( 2013 ) .
houck , h.e .
treci , and j. koch , nat .
phys . 8 , 292 - 299 ( 2012 ) .
m. hofheinz , h. wang , m. ansmann , r.c .
bialczak , e. lucero , m. neeley , a. d. oconnell , d. sank , j. wenner , j.m .
martinis , and a. n. cleland , nature 459 , 546 - 549 ( 2008 ) .
su , c .- p .
yang , s .- b .
zheng , scientific reports 4 , 3898 ( 2014 ) .
obrien , a. furusawa , j. vuckovi , nat .
phot . 3 , 687 - 695 ( 2009 ) . g. kurizki , p. bertet , y. kubo , k. mlmer , d. petrosyan , p. rabl , and j. schmiedmayer , pnas 2015 112 ( 13 ) 3866 - 3873 ( 2015 ) .
r. blatt and d.j .
wineland , nature 453 , 100815 ( 2008 ) .
d. nigg , m. mller , e. a. martinez , p. schindler , m. hennrich , t. monz , m. a. martin - delgado , and r. blatt , science 345 ( 69194 ) 302 - 305 ( 2014 ) . c. brif , r. chakrabarti , and h. rabitz , new journal of physics 12 , 075008 ( 2010 ) .
a.i . konnov and v.f .
automation and remote control , vol .
60 , no . 10 ( 1999 ) .
n. khaneja , t. reiss , c. kehlet , t. schulte - herbrggen , s. j. glaser , j. magn . reson .
172 , 296 - 305 ( 2005 ) .
d. dalessandro , introduction to quantum control and dynamics , taylor & francis ltd : hoboken , nj ( 1996 ) .
herschel a. rabitz et al .
science 303 , 1998 ( 2004 ) .
t .- s . ho , h. rabitz , j. photochem .
a 180 , 226 - 240 ( 2006 ) .
p. de fouquieres , s. schirmer , infin .
16 , 1350021 ( 2013 ) . k. w. moore , r. chakrabarti , g. riviello , h. rabitz , phys .
a 83 , 012326 ( 2011 ) .
r. b. wu , r. long , j. dominy , t. ho , and h. rabitz , phys .
a 86 , 013405 ( 2012 ) .
g. riviello , c. brif , r. long , r .- b .
wu , k. moore tibbetts , t. s. ho , h. rabitz , phys .
a 90 , 013404 ( 2014 ) .
alexander n. pechen and david j. tannor , phys .
106 , 120402 ( 2011 ) .
h. rabitz , t .- s . ho , r. long , r. wu , and c. brif .
108 , 198901 ( 2012 ) .
a. n. pechen , d. j. tannor .
phys . rev .
108 , 198902 ( 2012 ) .
a. n. pechen , n. ilin , phys .
a 86 , 052117 ( 2012 ) .
a. n. pechen , d. j. tannor , canadian journal of chemistry 92 , 157 - 159 ( 2014 ) .
t. caneva , t. calarco , and s. montangero , new j. phys .
14 , 093041 ( 2012 ) .
p. watts , j. vala , m. m. mller , t. calarco , k. b. whaley , d. m. reich , m. h. goerz , c. p. koch .
a 91 , 062306 ( 2015 ) .
g. gualdi , and d. m. reich and c. p. koch , and f. motzoi and k. b. whaley , and j. vala , and m. m. mller , s. montangero and t. calarco .
a 91 , 062307 ( 2015 ) .
p. doria , t. calarco , and s. montangero , phys . rev
106 , 190501 ( 2011 ) .
t. caneva , t. calarco , and s. montangero , phys .
a 84 , 022326 ( 2011 ) .
t. caneva , a. silva , r. fazio , s. lloyd , t. calarco , s. montangero , phys .
rev . a 89 , 042322 ( 2014 ) . m. m. mller , a. klle , r. lw , t. pfau , t. calarco , and s. montangero . phys .
a 87 , 053412 ( 2013 ) .
j. scheuer , x. kong , r. s. said , j. chen , a. kurz , l. marseglia , j. du , p. r. hemmer , s. montangero , t. calarco , b. naydenov , and f. jelezko , new j. phys .
16 , 093022 ( 2014 ) .
s. rosi , a. bernard , n. fabbri , l. fallani , c. fort , m. inguscio , t. calarco , and s. montangero , phys .
a 88 , 021601(r ) ( 2013 ) .
s. van frank , a. negretti , t. berrada , r. bcker , s. montangero , j .- f .
schaff , t. schumm , t. calarco , and j. schmiedmayer .
comm . 5 , 4009 ( 2014 ) . s. lloyd , s. montangero .
113 , 010502 ( 2014 ) .
moore , h. rabitz , j. chem .
137 , 134113 ( 2012 ) .
this is inspired but not to be confused with `` dressed crab '' , i.e. a dish where the crab meat is presented in the crab s cleaned shell to enjoy the meat without any effort .
powell , the computer journal 7 , 155 ( 1964 ) .
j. a. nelder and r. mead , computer journal 7 , 308 ( 1965 ) .
g. riviello , k. moore tibbetts , c. brif , r. long , r .- b .
wu , t .- s .
ho , h. rabitz , arxiv:1502.00707 ( 2015 ) . m. hsieh , t.s . ho , and h. rabitz , chem .
352 , 77 - 84 ( 2008 ) . to see this assume @xmath146 .
then by the regularity of the point there is a pulse update @xmath37 so that the fidelity @xmath147 is improved by @xmath37 as @xmath148 , i.e. @xmath149 and thus @xmath86 . if instead @xmath89 we have @xmath150 , i.e. @xmath151 . c. altafini , j. math .
phys , 43(5 ) , 2051 - 2062 ( 2002 ) .
this work was performed on the computational resource bwunicluster funded by the ministry of science , research and arts and the universities of the state of baden - wrttemberg , germany , within the framework program bwhpc . | in quantum optimal control theory the success of an optimization algorithm is highly influenced by how the figure of merit to be optimized behaves as a function of the control field , i.e. by the control landscape .
constraints on the control field introduce local minima in the landscape false traps which might prevent an efficient solution of the optimal control problem .
rabitz et al .
[ science 303 , 1998 ( 2004 ) ] showed that local minima occur only rarely for unconstrained optimization . here , we extend this result to the case of bandwidth - limited control pulses showing that in this case one can eliminate the false traps arising from the constraint .
based on this theoretical understanding , we modify the chopped random basis ( crab ) optimal control algorithm and show that this development exploits the advantages of both ( unconstrained ) gradient algorithms and of truncated basis methods , allowing to always follow the gradient of the unconstrained landscape by bandwidth - limited control functions .
we study the effects of additional constraints and show that for reasonable constraints the convergence properties are still maintained .
finally , we numerically show that this approach saturates the theoretical bound on the minimal bandwidth of the control needed to optimally drive the system .
the ability of achieving a desired transformation of a quantum system lies at the heart of the success of experiments in cold atoms @xcite , quantum optics @xcite , condensed matter @xcite , and in quantum technologies @xcite .
together with the fast development of these fields and the increasing complexity of the experiments , to develop efficient protocols it is often necessary to automatize the optimization process if not the whole development of the experimental sequence @xcite .
one possible way to perform such an optimization is by means of quantum optimal control , an approach that has proven to be very successful in solving this class of problems @xcite .
the success of gradient methods to find global solutions in quantum control problems @xcite is largely due to the fact that local minima are very rare in the control landscapes of a large class of systems @xcite .
however , to face the new challenges posed by the recent advancements in quantum science , gradient methods might be not the best option since to numerically calculate the gradient of the control objective might be quite inefficient . moreover in an increasing number of interesting applications , the control objective does not allow for an analytical calculation of the gradient @xcite .
the crab optimal control algorithm operates with an expansion of the control field onto a truncated basis and a direct optimization of the coefficients of the expansion by means of gradient - free minimization @xcite .
these characteristics allow for the solution of optimal control problems involving many - body quantum systems @xcite , as well as in the presence of long - range interactions @xcite , bandwidth - limited control according to experimental constraints @xcite , and highly nonlinear functionals where the gradient can not be calculated @xcite . however , the crab optimization does not necessarily fulfill the condition under which local minima in the optimization occur only rarely as it is by construction bandwidth - limited @xcite .
in this paper we extend the results presented by rabitz and coworkers @xcite and show that bandwidth - limited optimal control can be made virtually free of local minima as in the case of unconstrained control .
moreover , we numerically show that the global minima one always reaches correspond to an optimal solution if the bandwidth satisfy the theoretical bound given in reference @xcite . in particular , we present an extension of the crab algorithm the `` dressed crab '' ( dcrab ) that keeps the benefits of the original algorithm and comes with the additional property of guaranteed convergence to the global optimum in the cases where this is guaranteed also for gradient methods .
we test the two versions of the algorithm by optimizing a state transfer for different instances of a random spin hamiltonian .
being @xmath0 the dimension of the hilbert space the problem is defined on , for the standard crab algorithm false traps occur when the heuristic `` @xmath1-rule '' for the number of required control coefficients is violated @xcite . on the contrary , using the dcrab presented hereafter , all false traps are removed regardless of the number of control coefficients , resulting also in a faster convergence to the global optimum .
we present a theoretical explanation supporting these findings and showing that one can construct a set of random basis functions which follow the instantaneous gradient of the control landscape .
finally , we examine the behavior of dcrab in the presence of additional constraints on the bandwidth and amplitude of the pulse , and show that the algorithm is well - behaving as false traps appear only in the presence of strong constraints . in the following , for simplicity we will focus on state to state transfer of pure states , however the theoretical arguments are valid in the general scenario of unitary gate generation and mixed - states optimal control .
the structure of this paper is as follows : we first review the standard crab algorithm and introduce its extension , dcrab @xcite . in section
ii we present the theoretical basis of dcrab , while in section iii we show that dcrab follows the instantaneous gradient of the landscape .
finally , in section iv we apply both crab and dcrab to control problems with random spin hamiltonians and evaluate their performance with respect to success probability , computational effort and behavior under constraints . |
the rhic program at bnl has shown that the azimuthal asymmetry in momentum space , namely the elliptic flow @xmath3 , is the largest ever seen in hic suggesting that an almost perfect fluid with a very small shear viscosity to entropy density ratio , @xmath0 , has been created @xcite .
nonetheless it has been found that dissipative effects can not be neglected and even a small shear viscosity to entropy ratio @xmath0 produce sizeable effect increasing with the transverse momentum @xmath1 of the particles @xcite .
this has triggered a lot of activity in developing a relativistic theory of viscous hydrodynamics .
viscous corrections to ideal hydrodynamics are indeed large and a simple relativistic extension of first order navier stokes equations is affected by causality and stability pathologies @xcite .
it is therefore necessary to go to second order gradient expansion , and in particular the israel - stewart theory has been implemented to simulate the rhic collisions providing an upper bound for @xmath4 @xcite . such an approach , apart from the limitation to 2 + 1d simulations , has the more fundamental problem of a limited range of validity in @xmath0 and in the transverse momentum @xmath1
. on the contrary a relativistic transport approach has the advantage to be a 3 + 1d approach not based on a gradient expansion in viscosity that is valid also for large @xmath0 and for out of equilibrium momentum distribution allowing a reliable description also of the intermediate @xmath1 range where the important property of quark number scaling ( qns ) of @xmath5 has been observed @xcite . in this @xmath1 region
viscous hydrodynamics breaks its validity because the relative deviation of the equilibrium distribution function @xmath6 increases with @xmath7 becoming large already at @xmath8 @xcite .
furthermore boltzmann - vlasov transport theory distinguishes between the short range interaction associated to collisions and long range interaction associated to the field interaction , responsible for the change of the equation of state ( eos ) respect to that of a free gas .
this last feature allows to unify two main ingredients that are relevant for the formation of collective flow . in ideal hydrodynamics
the @xmath9 depends strongly on the eos namely on the sound velocity @xmath10 @xcite , while the mean free path @xmath11 is assumed to be vanishing . in the parton cascade approach the eos
is fixed to be the one of a free gas @xmath12 , on the other hand the mean free path @xmath13 is finite . in the first stage of rhic
the two different approaches were able to account for the large @xmath9 observed ; in particular the parton cascade with large scattering cross section predicted the saturation of @xmath9 vs @xmath1 @xcite .
anyway once viscosity is finite both a finite @xmath11 and the eos are important for the generation of the momentum anisotropies .
the basic equation of transport for the ( anti- ) quark phase - space distribution function @xmath14 under a scalar field interaction can be written as : @xmath15 where the first term is related to the free streaming , the second term represents the effect of a scalar field modifying the @xmath16 relation ( giving a finite interaction measure , see fig.2 ( left ) ) and the last term is the effect of the collisions directly associated to a finite @xmath11 and therefore to a finite @xmath0 . in the following
we will show two results obtained with such an approach : one related with the large @xmath2 ratio observed at rhic and the other discussing the role of the njl chiral dynamics on the relation between @xmath9 and @xmath0 .
we will show calculations for @xmath17 at 200 agev ; the density distribution in coordinate space is given by the standard glauber model .
the maximum initial temperature is @xmath18 mev and the initial time is @xmath19 fm / c as usually done also in hydrodynamical calculations .
we first consider the reduction of the transport approach to a cascade , i.e. neglect the field interaction , included in the next section .
in such a case the mass @xmath20 in eq.(1 ) . it has been shown that a fluid at finite @xmath0 can be studied within a transport approach renormalizing locally the scattering cross section according to @xmath21 , for details see ref.s @xcite .
the system of course expands and cools and when the energy density is below @xmath22 it starts the cross - over region and then the hadronic stage . during this phase we increase the @xmath23 of the medium according to fig . 1 ( left ) in order to have finally in the hadronic phase an @xmath24 in agreement with several calculation of @xmath0 . at rhic
it has been observed @xmath25 ratio of about 0.8 - 0.9 quite larger than first ideal hydrodynamical predictions of about 0.5 .
more realistic hydrodynamical calculations show that the ratio can be even lower than 0.5 and viscosity does not strongly affect such a ratio but can bring it up to about 0.6 . furthermore in ref.@xcite
it has been shown that such a ratio is almost independent on the initial eccentricity , on the viscosity , on the particle species .
therefore even if @xmath26 and @xmath3 can depend on several details such a ratio is a quite stable one and its large value should be considered as an oppurtunity to have some new information on the system not yet spot . on the energy density @xmath27 implemented in the transport code .
right : @xmath2 for @xmath17 collisions in the mid - rapidity region @xmath28 at b= 8 fm ; dashed line is the result for a constant @xmath29 , the solid line is the result for a temperature dependence of the @xmath0 like shown by solid line on the left figure.,title="fig:",width=240,height=172 ] on the energy density @xmath27 implemented in the transport code .
right : @xmath2 for @xmath17 collisions in the mid - rapidity region @xmath28 at b= 8 fm ; dashed line is the result for a constant @xmath29 , the solid line is the result for a temperature dependence of the @xmath0 like shown by solid line on the left figure.,title="fig:",width=240,height=192 ] in ref.@xcite we have shown that indeed the formation time of @xmath26 is larger than the @xmath3 one and therefore it is more affected by dynamics of the later stage of the reaction . thanks to the difference in the formation time we argue that such a ratio is significantly affected by the temperature dependence of the @xmath0 .
therefore we have performed two calculations one with a constant @xmath29 in the qgp phase ( dashed line ) and the other , shown by solid line in fig . 1 ( left ) , with a t or @xmath27 energy density dependence @xmath30 where @xmath31 gives the beginning of the cross over region .
we remind that an increase of @xmath0 is expected moving away from the phase transition region on general considerations @xcite . in fig.1 ( left )
we show the results of the two calculations for @xmath17 at 200 agev . for a constant @xmath0
we find a value of about 0.6 while in the case of a temperature dependent @xmath0 , it increases getting very close to experimental data , @xmath32 0.8 .
the reason for the sensitivity of this ratio on the t dependence is that the elliptic flow @xmath3 mainly develops earlier respect to @xmath26 and therefore on the average at larger @xmath0 ( for the solid line case ) , while @xmath26 develops more when the @xmath0 is smaller hence is less suppressed by dissipative effects .
to start the investigation of the role of the field interaction we have implemented the njl dynamics that introduces the generation of a finite mass associated to chiral symmetry breaking .
therefore particles do not have a vanishing mass but a finite one according to the mass gap equation of the njl model : @xmath33 that determines the local mass @xmath34 at the space - time point @xmath35 in terms of the distribution functions @xmath36 . for a gas at finite mass @xmath37
is no longer vanishing as shown in fig.2 ( right ) for several njl parametrization .
eqs.([vlasovnjl ] ) and ( [ noneqmass2 ] ) form a closed system of equations constituting the boltzmann - vlasov equation associated to the njl lagrangian .
the space - time dependence of the mass @xmath38 influences the momenta of the particles because the finite gradient of the condensate generates a force which changes the momentum of a particle proportionally to @xmath39 .
the last is negative because the phase transition occurs earlier in the surface of the expanding qgp fireball .
therefore the phase transition which take place locally results in a negative contribution to the particle momenta that makes the system more sticky respect to a free massless gas . to study a fluid at finite @xmath0 and njl field interaction one has to extend the simple relation @xmath40 valid for a massless gas to the general case of massive relativistic particles and
this has been discussed in ref.@xcite . in fig.[v2-pt ] , we show the elliptic flow as a function of @xmath1 .
we find that the presence of an njl - field that drives the chiral phase transition suppress the @xmath5 by about @xmath41 at @xmath42 gev .
this would imply the need of a parton scattering cross section @xmath43 even larger than that estimated with the cascade model which was already quite larger than the pqcd estimates @xcite in order to describe the data .
on the other hand the mean field modifies both the local entropy density , reduced by the mass generation , and the shear viscosity that increases respect to the massless case . considering that one of the main goal is to determine the @xmath23 of the qgp , we have investigated what is the action of the mean field once the @xmath23 of the system is fixed to be the same with and without the njl field .
this of course implies different values of the cross section in the two cases .
collisions in the mid - rapidity region @xmath28 at b= 7 fm ; see text.,title="fig:",width=192,height=172 ] collisions in the mid - rapidity region @xmath28 at b= 7 fm ; see text.,title="fig:",width=249,height=172 ] the results for @xmath44 are shown by dashed lines in fig .
[ v2-pt ] ( right ) , the ( black ) dashed line is the case with only the collision term ( cascade ) while the ( green ) dashed line is the case with the field .
we can see that once the @xmath23 is fixed there is essentially no difference in the calculations with and without a field dynamics included .
this is a key result that shows that even in a microscopic approach that distinguishes between the mean field and the collisional dynamics the @xmath5 is mainly driven by the @xmath23 of the fluid . in other words we have found that in a microscopic approach the @xmath23 is the pertinent parameter and the language of viscous hydrodynamics is appropriate .
of course this does not mean that @xmath5 in the transport theory is the same as in viscous hydrodynamics ( especially at @xmath45 gev ) , but that the direct relation between @xmath5 and @xmath23 is not modified by the njl field dynamics .
this confirms the validity of the studies pursued till now even if they miss an explicit mean field dynamics and/or the chiral phase transition .
this work for v. baran is supported in part by the romanian ministry for education and research under the cncsis contract pnii id-946/2007 . | we introduce a transport approach at fixed shear viscosity to entropy ratio @xmath0 to study the generation of collective flows in ultra - relativistic heavy - ion collisions .
transport theory supplies a covariant approach valid also at large @xmath0 and at intermediate transverse momentum @xmath1 , where deviations from equilibrium is no longer negligible .
such an approach shows that at rhic energies a temperature dependent @xmath0 enhances significantly the @xmath2 respect to the case of constant @xmath0 .
furthermore if njl chiral dynamics is self - consistently implemented we show that it does not modify the relation between @xmath3 and @xmath0 . |
multi - component quantum hall systems have exhibited a collection of interesting phenomena which are the manifestation of electronic correlations @xcite .
such correlations become particularly prominent when two or more sets of landau levels ( lls ) with different layer , subband , valley , spin , or landau level indices are brought into degeneracy @xcite . in experimental systems , different lls can be tuned to cross by varying gate voltage , charge density , magnetic field or the magnetic field tilted angle to the sample .
one of the attractions is the formation of quantum hall ferromagnets ( qhfs ) due to the exchange interactions of the two subbands states , termed as pseudospins @xcite .
self - consistent local density approximation ( sclda ) and hartree - fock mean field method can be performed on the calculation of ground state energy and quasi - particle energy gap @xcite .
recent experiments in single quantum well with two - subband occupied systems @xcite , showed evidence of qhfs when two lls were brought into degeneracy .
the qhfs can either be easy - axis or easy - plane , depending on the details of ll crossing configurations @xcite . in this paper
we follow the theoretical framework of jungwirth and macdonald @xcite and numerically calculate the pseudospin anisotropy energy using the sample parameters in the experiment of zhang _ et al .
the result confirms the qhfs taking place at total filling factor @xmath2 and @xmath3 are expected to be easy - plane and easy - axis qhfs , respectively .
as we will discuss in this paper , the gate bias voltage , which affects the spatial distributions of the wave functions of the two subbands , plays a leading role in the formation of easy - plane and easy - axis qhfs .
the pseudospin representation is used to describe the valence lls degenerated in the fermi level @xcite . following the theoretic studies of pseudospin qhfs @xcite ,
here we focus on a two - subband two dimensional electron system , in which the lls are labeled by @xmath4 , where @xmath5 is first / second subband ( in the no biased quantum well , also called symmetry / antisymmetry subband ) index , @xmath6 is landau level in - plane orbit radius quantum number , and @xmath7 represents real spin .
when two lls are brought close to degeneracy but still sufficiently far from other lls , one of them can be labeled as pseudospin up ( @xmath8 ) and the other as pseudospin down ( @xmath9 ) . in the experimental work of zhang _
@xcite , as shown in fig .
[ fig : landau ] , we can label pseudospin up ( @xmath8 ) as @xmath10 and pseudospin down ( @xmath9 ) as @xmath11 at filling factor @xmath12 . at filling factor @xmath3
, we label pseudospin up ( @xmath8 ) as @xmath13 and pseudospin down ( @xmath9 ) as @xmath14 . here the upper and lower signs before spin index refer to the crossing point at lower and higher magnetic field , respectively .
are crossing to each other during the increasing of the magnetic field ( see text for details ) .
three arrows point out the crossing point of landau levels at filling factor @xmath15 and @xmath16 . ] at this moment , we would emphasize the essential similarity between two - subband system and ordinary bilayer system ( double quantum well ) . on one hand , due to a separation of charges to opposite sides of the well originating from the coulomb repulsion , a wide single quantum well can be modeled as an effective bilayer system wescheider , macdonald2000,abolfath .
therefore in a single quantum well two - subband structure , the states of electrons can be characterized by two parameters similar to a bilayer system : the tunneling gap @xmath17 and effective layer separation @xmath18 .
the @xmath17 is chosen as equal to the difference between the lowest two subband energy levels and @xmath18 is given by the distance between two centers of the charge distribution in this single quantum well .
on the other hand , let us assume the two dimensional electron gas is in the @xmath19 plane in the following calculation and discussion , so the sample growth direction is aligned with the @xmath20 axis . in a no biased bilayer system , although the layer wave function in @xmath20 direction is spatially separated in two layers , once included in the effect of tunneling between two layers @xmath21 and rediagonalized the @xmath20 direction hamiltonian , it will be just like the two - subband system which has symmetry and antisymmetry subbands . and the symmetry - antisymmetry gap @xmath22 is equal to the inter - layer tunneling @xmath21 . even when a bias voltage between two layers exists , the circumstance is more or less the same .
since they are theoretically identical , all the pseudospin language theory used in bilayer system is available in the two - subband system .
while pseudospin up and pseudospin down lls are degenerate but the number of electrons is not enough to fill all the two lls , electrons will stay in a broken - symmetry ground state .
actually , the state electrons choosing is a linear combination of two pseudospin lls which minimizes the system total energy .
typically , the many - body ground state can be written as follows macdonald2000 : @xmath23 where @xmath24 creates the single - particle state oriented in a certain unit vector @xmath25 with wave function : @xmath26 here @xmath27 is the single - particle state wave function which contains growth direction subband wave function @xmath28 and in - plane ll wave function @xmath29 : @xmath30 then the hartree - fock energy of the system can be obtained as following macdonald2000 : @xmath31|h|\psi \lbrack \hat{m}]\rangle } { n_{\phi } } = -\sum_{i = x , y , z}\left ( e_{i}-\frac{1}{2}u_{% \mathbf{1},i}-\frac{1}{2}u_{i,\mathbf{1}}\right ) m_{i}+\frac{1}{2}% \sum_{i , j = x , y , z}u_{i , j}m_{i}m_{j}. \label{eq : hf_energy}\]]whether the ground state is easy - axis or easy - plane only depends on the quadratic coefficient in pseudospin magnetization @xmath32 ( i.e. the pseudospin anisotropy energy @xmath33 , @xmath34 and @xmath35 ): @xmath36the first @xmath37 term in eq .
( [ eq : anisotropy_energy ] ) is the hartree term and the second term is the fock term .
@xmath38 can be expanded to sum of several pseudospin matrix elements @xmath39 , which are products of subband and the in - plane parts @xcite . if @xmath40 , the system is in easy - axis qhf , which means the pseudospin magnetization @xmath41 is aligned either up or down ; and if @xmath42 , the system is in easy - plane qhf , which is a coherent superposition of the two pseudospin lls .
we numerically calculate the growth direction ( @xmath20 direction ) wave function @xmath28 in eq .
( [ eq : single - particle ] ) using sclda to compare the pseudospin anisotropy energy @xmath33 and @xmath43 in eq .
( [ eq : hf_energy ] ) @xcite . in the sclda method , wave function @xmath44
is described by the schrdinger equation : @xmath45 here @xmath46 is the effective electron mass in gaas , @xmath47 corresponds to the conduction band discontinuity , @xmath48 is the bias potential caused by the difference of front and back gate voltage @xmath49 , @xmath50 refers to the exchange - correlation potential related to the electron charge distribution @xmath51 ( we use the form of @xmath52 given by hedin and lundqvist @xcite ) .
the hartree term @xmath53 due to electrostatic potential is given in the poisson equation : @xmath54 the subband energies @xmath55 , wave functions @xmath28 and the electron charge distributions @xmath56 of both subbands can be calculated by solving the schrdinger equation eq .
( [ eq : schrodinger ] ) and the poisson equation eq .
( [ eq : poisson ] ) simultaneously algorithm .
taking the unit of energy as @xmath57 ( @xmath58 is the magnetic length ) , we calculate the pseudospin anisotropy energy @xmath33 , @xmath34 and @xmath35 at filling factor @xmath59 and @xmath60 in zhang _
et al . _ s work @xcite . at total filling factor @xmath59 , @xmath61 , so the system will stay in easy - plane qhf . at one degenerate point of total filling factor
@xmath60 , @xmath62 , and at the other degenerate point @xmath63 .
both of the energy differences of @xmath64 at @xmath60 indicate easy - axis qhf ground states .
to illustrate the evolution from easy - axis qhf to easy - plane qhf , we calculate the phase diagrams of @xmath64 as a function of bias gate voltage @xmath65 and total density @xmath66 at filling factor @xmath59 ( fig . [ fig : v3_phase ] ) and @xmath60 ( fig . [ fig : v4_phase ] ) . in the following discussion , in order to make consistency , we choose @xmath67 ( @xmath68 ) as the unit of energy . in the phase diagram ( fig .
[ fig : phase ] ) , we label the density and bias voltage @xmath69 position where the crossing occurs in the experimental work of zhang _
_ @xcite . in the left / blue ( right / red ) parts of each figure ( fig .
[ fig : v3_phase ] and fig .
[ fig : v4_phase ] ) , where @xmath70 , easy - axis ( easy - plane ) has the lower electron interaction energy . from fig .
[ fig : phase ] , we find that the anisotropy energy difference @xmath64 is very sensitive to the bias voltage @xmath65 .
if we could vary the gate voltage across the black line labeled @xmath71 from left to right in a determined density , a quantum phase transition from easy - axis to easy - plane qhf will happen . and @xmath35 v.s . bias voltage @xmath72 .
except for @xmath73 at @xmath74 , all the other terms are same at filling factor @xmath15 and @xmath75 .
@xmath76 and @xmath77 correspond to the hartree term and fock term in the hartree - fock calculation of @xmath35 . ] in order to make a comparison of the effect of each term in anisotropy energy , we plot them in eq .
( [ eq : anisotropy_energy ] ) as a function of @xmath69 at a certain density of @xmath78 in fig .
[ fig : energy ] .
here we have to point out that all the formulas at @xmath59 and @xmath79 are the same , except that @xmath33 vanishes at @xmath60 .
so the fig .
[ fig : energy ] is applicable to both filling factor @xmath59 and @xmath60 only if in the @xmath60 situation the @xmath80 term is set to a constant zero .
the reason for the different behavior of @xmath33 at filling factor @xmath59 and @xmath60 will be discussed in the following .
note that the dominated term in @xmath64 is the hartree term @xmath76 , which is due to the electrostatic potential of electrons .
it shows that the @xmath76 term increases immediately with the increasing bias voltage @xmath65 . as a function of bias gate voltage @xmath49 at the density of @xmath81 .
three insets show the well profiles ( black dashed lines ) with first ( blue solid lines ) and second ( red dash - dot lines ) subbands wave functions at different bias voltage : @xmath82 , @xmath83 , and @xmath84 . ] to examine what the role the bias voltage plays in the determination of easy - plane or easy - axis qhf , we give some sclda results in the fig .
fig : d_vg .
we define an effective subband separation @xmath85 to describe the distance between the cores of first and second subbands wave functions .
the @xmath86 as a function of bias voltage is plotted in fig .
[ fig : d_vg ] , also with some demonstrations of quantum well configurations and subbands wave functions in different bias voltages ( @xmath87 , @xmath88 , and @xmath89 ) . the data in fig .
[ fig : d_vg ] are all selected from the same density @xmath90 as in fig .
[ fig : energy ] .
it is obvious that the bias gate voltage changes the effective separation of two subbands while changing the well potential @xmath91 in eq .
( [ eq : schrodinger ] ) .
the larger the bias voltage is added , the farther the lowest two subbands are separated . since at a well separated @xmath20 direction wave function configuration , all the electrons near the fermi level filling in the same subband , i.e. the same pseudospin level , will raise a larger electrostatic energy , the easy - axis qhf is not favorable .
thus an easy - plane qhf can save more hartree energy in the larger bias voltage situation . on the other hand , when the potential of the quantum well maintains a good symmetry in the small bias voltage limit , electrons can pick up one of the two pseudospin levels to keep the hartree energy minimal and avoid the energetic penalty from inter - subband tunneling as well . as mentioned before
, the difference of @xmath33 between total filling factor @xmath59 and @xmath60 could be explained in the similar way .
the anisotropy energy @xmath33 or @xmath34 is constituted of hartree part and fock part ( see eq . ( [ eq : anisotropy_energy ] ) ) . in our numerical calculation , we find that the hartree terms of @xmath33 and @xmath34 are always zero at both @xmath59 and @xmath60 .
and the only non - zero term in @xmath33 or @xmath34 is the fock term @xmath92 at filling factor @xmath59 , which owes to the exchange interaction .
it implies that at filling factor @xmath60 , where pseudospin up @xmath93 and pseudospin down @xmath94 have opposite real spins , the easy - plane anisotropy , in which the electrons stay in both subband equally , would cost much more exchange energy .
but at @xmath59 the pseudospin up @xmath95 and pseudospin down @xmath96 have the same spin .
then there is no such problem need to be considered .
therefore , the easy - axis qhf is more likely to happen at total filling factor @xmath60 than @xmath59 .
on the basis of above discussion , we may summarize as following .
the bias gate voltage added to the sample changes the quantum well profile in the growth direction as well as the spacial separation of the lowest two subbands wave functions . for a larger bias gate voltage ,
the potential of the well is much skewer , so the two subbands wave functions locate in the opposite side of the quantum well ( inset of fig .
[ fig : d_vg ] ) . as a result ,
the hartree energy will arise if all the electrons stay in one narrow subband or pseudospin level .
thus a easy - plane qhf , in which the electrons fill the two pseudospin levels equally , is more energetically favorable at a large bias gate voltage .
in addition , a state with opposite real spins will expend more exchange energy , so the easy - plane qhf is more easily to form at a pseudospin configuration in which the two pseudospin level have the same real spin ( filling factor @xmath2 in this paper ) .
using self - consistent local density approximation method and hartree - fock mean field theory , we calculated wave function in the growth direction and the anisotropy energy in the two - subband quantum hall system .
the data shows great consistent with the observed easy - plane and easy - axis quantum hall ferromagnets at filling factor @xmath2 and @xmath3 in the experiment of zhang _
also , by analyzing the numerical result , we give an easy - to - understand explanation about the anisotropy occuring at these filling factors .
this work was funded by national fundamental research program , the innovation funds from chinese academy of sciences , ncet-04 - 0587 , and national natural science foundation of china ( grant no . 60121503 , 10574126 , 10604052 ) . for a review on quantum hall ferromagnets , see experimental chapter by j.p .
eisenstein and theoretical chapter by s. m. girvin and a. h. macdonald in s. das sarma and a. pinczuk ( eds . ) , _ perspectives on quantum hall effects _
( wiley , new york , 1997 ) . | we carry out a numerical study of the quantum hall ferromagnetism in a two - subband system using a set of experimental parameters in a recently experiment [ x. c. zhang , i. martin , and h. w. jiang , phys .
rev .
b * 74 * , 073301 ( 2006 ) ] . employing the self - consistence local density approximation for growth direction wave function and the hartree - fock theory for the pseudospin anisotropy energy ,
we are able to account for the easy - axis and easy - plane quantum hall ferromagnetism observed at total filling factor @xmath0 and @xmath1 , respectively .
our study provides some insight of how the anisotropy energy , which highly depends upon the distribution of growth direction wave functions , determines the symmetry of the quantum hall ferromagnets . |
recently , symmetric positive definite ( spd ) matrices of real numbers appear in many branches of computer vision .
examples include region covariance matrices for pedestrian detection @xcite and texture categorization @xcite , joint covariance descriptor for action recognition @xcite , diffusion tensors for dt image segmentation @xcite and image set based covariance matrix for video face recognition @xcite . due to the effectiveness of measuring data variations ,
such spd features have been shown to provide powerful representations for images and videos .
however , such advantages of the spd matrices often accompany with the challenge of their non - euclidean data structure which underlies a specific riemannian manifold @xcite . applying the euclidean geometry directly to spd matrices
often results in poor performances and undesirable effects , such as the swelling of diffusion tensors in the case of spd matrices @xcite . to overcome the drawbacks of the euclidean representation , recent works @xcite have introduced riemannian metrics , e.g. , affine - invariant metric @xcite , log - euclidean metric @xcite , to encode the riemannian geometry of spd manifold properly . by applying these classical riemannian metrics ,
a couple of works attempt to extend euclidean algorithms to work on manifolds of spd matrices for learning more discriminative spd matrices or their vector - forms . to this end ,
several studies exploit effective methods on one spd manifold by either flattening it via tangent space approximation @xcite ( see fig.[fig1 ] ( a)@xmath0(b ) ) or mapping it into a high dimensional reproducing kernel hilbert space ( rkhs ) @xcite ( see fig.[fig1 ] ( a)@xmath0(c)@xmath0(b ) ) .
obviously , both of the two families of methods inevitably distort the geometrical structure of the original spd manifold due to the procedure of mapping the manifold into a flat euclidean space or a high dimensional rkhs .
therefore , the two learning schemes would lead to sub - optimal solutions for the problem of discriminative spd matrix learning .
( b ) is to firstly flatten the original manifold @xmath1 by tangent space approximation and then learn a map @xmath2 to a discriminative euclidean space @xmath3 .
the second one ( a)@xmath0(c)@xmath0(b ) is to firstly embed @xmath1 with an implicit map @xmath4 into an rkhs @xmath5 and then learn a mapping @xmath6 to a more discriminative euclidean space @xmath3 .
the last one ( a)@xmath0(d ) aims to learn a map @xmath7 from the original spd manifold @xmath1 to a more discriminative spd manifold @xmath8 . here , @xmath9 and @xmath10 are the spd matrices , @xmath11 and @xmath12 are the tangent spaces . ] [ fig1 ] to more faithfully respect the original riemannian geometry
, another kind of spd - based discriminant learning methods @xcite aims to pursue a column full - rank transformation matrix mapping the original spd manifold to a more discriminative spd manifold , as shown in fig.[fig1 ] ( a)@xmath0(d ) .
however , as directly learning the manifold - manifold transformation matrix is hard , the work @xcite alternatively decomposes it to the product of an orthonormal matrix with a matrix in gl@xmath13 , and requires the employed riemannian metrics to be affine invariant . by doing so ,
optimizing the manifold - manifold transformations is equivalent to optimizing over orthonormal projections .
although the additional requirement simplifies the optimization of the transformation , this has not only reduced the original solution space but also inevitably excluded all non - affine invariant riemannian metrics such as the well - known log - euclidean metric , which has proved to be much more efficient than affine - invariant metric @xcite .
while the work @xcite exploited the log - euclidean metric under the same scheme , it actually attempts to learn a tangent map , which implicitly approximate the tangent space and hence introduces some distortions of the true geometry of spd manifolds . in this paper , also under the last scheme ( see fig.[fig1 ] ( a)@xmath0(d ) ) , we propose a new geometry - aware spd similarity learning ( spdsl ) framework to open a broader problem domain of learning discriminative spd features by exploiting either affine invariant or non - affine invariant riemannian metrics on spd manifolds . to realize the spdsl framework , there are three main contributions in this work : * by exploiting the riemannian geometry of the manifold of fixed - rank positive semidefinite ( psd ) matrices , our spdsl framework provides a new solution to directly learn the manifold - manifold transformation matrix . as no additional constraint
is required , the optimal transformation will be pursued in a favorable solution space , enabling a wide range of well - established riemannian metrics to work as well . * to fulfill the solution , a new supervised spd similarity learning technique is proposed to learn the transformation by regressing the similarities of selected spd pairs to the target similarities on the resulting spd manifold .
* we derive an optimization approach which exploits the classical riemannian conjugate gradient ( rcg ) algorithm on the psd manifold to optimize the proposed objective function .
let @xmath14 be a set of real , symmetric matrices of size @xmath15 and @xmath16 be a set of spd matrices .
the mapping space @xmath17 is endowed with usual euclidean metric ( i.e. , inner product ) @xmath18 . as noted in @xcite , the set of spd matrices @xmath19 is an open convex subset of @xmath17 .
thus , the tangent space to @xmath19 at any spd matrix in it can be identified with the set @xmath17 .
a smoothly - varying family of inner products on each tangent space is known as riemannian metric , endowing which the space of spd matrices @xmath19 would yield a riemannian manifold . with such riemannian metric , the geodesic distance between two elements @xmath20 on the spd manifold is generally measured by @xmath21 . several riemannian metrics and divergences have been proposed to equip spd manifolds . for example
, affine - invariant metric @xcite , stein divergence @xcite , jeffereys divergence @xcite are designed to be invariant to affine transformation .
that is , for any @xmath22 ( i.e. , the group of real invertible @xmath23 matrices ) , the metric function @xmath24 has the property @xmath25 .
in contrast , log - euclidean metric@xcite , cholesky distance @xcite and power - euclidean metric @xcite are not affine invariant . among these metrics , only affine - invariant metric @xcite and log - euclidean metric @xcite
define a true geodesic distance on the spd manifold @xcite .
in addition , the stein divergence are also widely used due to its favorable properties and high performances in visual recognition tasks @xcite .
therefore , this paper focuses on studying such three representative riemannian metrics .
* definition 1 . * _ by defining the inner product in the tangent space at the spd point @xmath26 on the spd manifold as @xmath27 and the logarithmic maps as @xmath28 , the geodesic distance between two spd matrices @xmath20 on the spd manifold is induced by affine - invariant metric ( aim ) as _
* definition 2 . * _ the approximated geodesic distance between two spd matrices @xmath20 on the spd manifold is defined by using stein divergence as _
* definition 3 . * _ by defining the inner product in the space at the spd point @xmath26 on the spd manifold as @xmath31 , \text{d}\log(\bm{x}_1)[\bm{h}_2 ] \rangle$ ] ( @xmath32 $ ] denotes the directional derivative ) and the logarithmic maps as @xmath33 $ ] , the geodesic distance between two spd matrices @xmath20 is derived by log - euclidean metric ( lem ) as _
in this section , we first propose a new solution of riemannian geometry - aware dimensionality reduction for spd matrices , and then present our supervised spd similarity learning method under the solution .
finally , we give a detailed description of our developed optimization algorithm .
given a set of spd matrices @xmath35 , where each matrix @xmath36 , and a transformation @xmath37 ( @xmath38 ) is pursued for mapping the original spd manifold @xmath19 to a lower - dimensional spd manifold @xmath39 .
formally , this procedure attempts to learn the parameter @xmath40 , of a mapping in the form @xmath41 , which is defined as : @xmath42 to ensure the resulting mapping yields a valid spd manifold @xmath43 @xmath44 , the manifold - manifold transformation @xmath45 is basically required to be a full - rank matrix @xmath46 .
since the solution space is a non - compact stiefel manifold @xmath47 where the distance function has no upper bound , directly optimizing on the manifold is infeasible .
fortunately , the conjugates ( taking the form of @xmath48 ) of column full - rank matrices span a compact manifold @xmath49 of positive semidefinite ( psd ) matrices , which is a quotient space of @xmath47 and owns a well - established riemannian structure .
in contrast , by additionally assuming the transformation @xmath40 to be orthogonal as done in @xcite , eqn.[eq3 ] could be optimized on compact stiefel manifold , which is a subset of the non - compact stiefel manifold @xmath47 .
further , for the affine invariant metrics ( e.g. , aim ) , optimizing on stiefel manifold can be reduced to optimizing over grassmannian @xcite .
however , such orthogonal solution space is smaller than the original solution space @xmath47 , making the optimization theoretically yield suboptimal solution of @xmath40 .
thus , we choose to optimization on the psd manifold to search the optimal solution of @xmath40 in a more faithful way .
now , we need to study the geometry of the psd manifold @xmath49 . for all orthogonal matrices @xmath50 of size @xmath51 , the map @xmath52 leaves @xmath48 unchanged .
this property of @xmath40 results in the equivalence class of the form @xmath53 = \{\bm{wo}| \bm{o } \in \mathbb{r}^{m \times m } , \bm{o}^t\bm{o}=\bm{i}_m\}$ ] , and a one - to - one correspondence with the rank-@xmath54 psd matrix @xmath55 . by quotienting this equivalence relation out , the set of rank-@xmath54 psd matrices @xmath49
is reduced to the quotient of the manifold @xmath56 by the orthogonal group @xmath57 , i.e. , @xmath58 . with the studied relationship between @xmath49 and @xmath56 ,
the function @xmath59 is able to derive the function @xmath60 defined as @xmath61 . here ,
@xmath2 is defined in the total space @xmath56 and descends as a well - defined function in the quotient manifold @xmath49 .
therefore , optimizing over the total space @xmath56 is reduced to optimizing on the psd manifold @xmath49 , which is well - studied in several works @xcite .
note that , as each element @xmath62 on the psd manifold is simply parameterized by @xmath40 , optimizing on the manifold actually deals directly with @xmath40 .
to more easily understand this point , one can take the well - known grassmann manifold as an analogy , where each element can be similarly represented by the equivalence class @xmath53 $ ] or the projection matrix @xmath48 ( here , @xmath63 ) , and the optimization on it directly seeks the solution of @xmath40 .
it can be further proven that the quotient @xmath49 presents the structure of a riemannian manifold @xcite . as a result , endowing the total space @xmath56 with the usual riemannian structure of a euclidean space ( i.e. , the inner product @xmath18 ) , a riemannian structure for the quotient space @xmath49 follows .
the inner product occurs on the tangent space @xmath64 of the manifold @xmath56 . in the case of the manifold @xmath49
, the corresponding tangent space is decomposed into two orthogonal subspaces , the vertical space @xmath65 and the horizontal space @xmath66 , to achieve the inner product @xmath67 .
this riemannian metric facilitates several classical optimization techniques such as riemannian conjugate gradient ( rcg ) algorithm @xcite working on the psd manifold @xmath49 . as for more detailed background on the riemannian geometry of the psd manifold
, please refer to the works @xcite . by exploiting the riemannian geometry of the fixed - rank psd manifold @xmath49 , we here open up the possibility of directly pursuing an optimal column full - rank manifold - manifold transformation matrix to solve the problem of dimensionality reduction on spd features . as studied before , under the proposed framework of dimensionality reduction on spd features ,
a target spd manifold @xmath39 of lower dimensionality can be derived . on the new spd manifold @xmath39 ,
the geodesic distance between the two original spd points @xmath68 is achieved by : @xmath69 where @xmath70 is the manifold - manifold transformation computed by eqn.[eq3 ] , @xmath71 can be the geodesic distance induced by the commonly - used affine or non - affine invariant riemannian metrics eqn.[eq1 ] , eqn.[eq1.0 ] and eqn.[eq2 ] . in this paper
, we are focusing on the problem of supervised spd similarity learning for more robust visual classification tasks where spd features have shown great power .
formally , for each spd matrix @xmath36 , we define its class indicator vector : @xmath72 \in \mathbb{r}^c$ ] , where the @xmath73-th entry being 1 and other entries being 0 indicates that @xmath74 belongs to the @xmath73-th class of @xmath75 classes in total . as discriminant learning techniques developed in euclidean space
, we assume that prior knowledge is known regarding the distances between pairs of spd points on the new spd manifold @xmath39 .
let s take the similarity or dissimilarity between pairs of spd points into account : two spd points are similar if the similarity based on the geodesic distance between them on the new manifold is larger , while two spd points are dissimilar if their similarity is smaller . given a set of the similarity constraints , our goal is to learn the manifold - manifold transformation matrix @xmath40 that parameterizes the similarities of spd points on the target spd manifold @xmath39 . to this end
, we exploit the supervised criterion of centered kernel target alignment @xcite to learn discriminative features on the spd manifold by regressing the similarities of selected sample pairs to the target similarities .
formally , our supervised spd similarity learning ( spdsl ) approach is to maximize the following objective function : @xmath76 where @xmath77 and @xmath78 are frobenius inner product and norm respectively .
the elements of matrix @xmath79 encodes the similarities of spd data while the elements of @xmath80 presents the ground - truth similarities of the involved spd points .
the matrix @xmath81 is used to select the pairs of spd points when the corresponding elements are 1 .
the matrix @xmath82 is employed for centering the data similarity matrix @xmath79 and the similarity matrix @xmath80 on labels .
@xmath83 is the number of samples , @xmath84 is the identity matrix of size @xmath85 , @xmath86 is the vector of size @xmath83 with all entries being ones , @xmath87^t$ ] is here supposed to be centered , i.e. , @xmath88 , for simplicity . in the following , we will give the formulations of the two matrices @xmath89 and @xmath81 in more details .
more specifically , the employed matrix @xmath89 in eqn.[eq9 ] encodes the similarity between each pair of spd points @xmath90 on the spd manifold @xmath19 , which takes a form as : @xmath91 where @xmath92 is computed by eqn.[eq4 ] , @xmath93 is typically set as @xmath94 , @xmath95 is empirically set to mean of distances of the original training sample pairs .
actually , the function eqn.[eq6 ] takes a form of gaussian kernel function .
however , as the objective function eqn.[eq9 ] can be expressed as sum of the similarity regression results of selected sample pairs , the function eqn.[eq6 ] just serves as a tool to encode the similarities and is thus not necessarily positive definite ( pd ) . in practical application , the computational burden of handling the full kernel matrix @xmath89 on the spd
manifold scales quadratically with the size of training spd data . to address this problem
, we exploit the idea of graph embedding technique @xcite to select a limited number of data pairs to construct a sparse kernel matrix ( non pd ) with a large number of elements being zero . with this idea in mind ,
the matrix @xmath81 is defined to select the pairs of spd points for spd similarity learning . by employing it ,
@xmath96 can be regarded as the sparse kernel matrix , where the operation @xmath97 denotes hadamard product and the matrix @xmath98 . here ,
@xmath99 and @xmath100 are defined as : @xmath101 where @xmath102 is the set of @xmath103 nearest neighbors of @xmath74 that share the same class label as @xmath104 , and @xmath105 is the set of @xmath106 nearest neighbors of @xmath74 with different class labels from @xmath104 . according to the theory of graph embedding @xcite ,
the within - class similarity graph @xmath99 and the between - class dissimilarity graph @xmath100 respectively defined in eqn.[eq7 ] and eqn.[eq8 ] can encode the local geometrical structure of the space of the processing data .
thus , in addition to speeding up the discriminant learning on the spd features , exploiting the graph embedding technique can not only learn the discriminative information of spd data but also characterize the local riemannian geometry of the underlying spd manifold .
the efficiency and effectiveness of the proposed discriminant learning approach working on spd manifolds will be further studied in the experimental part . as discussed before , optimizing in the solution space @xmath56 of the column full - rank transformation matrices in our objective function
can be reduced to optimizing on the riemannian manifold of rank-@xmath54 psd matrices , @xmath49 .
therefore , in this section , we exploit the riemannian conjugate gradient ( rcg ) algorithm @xcite to optimize our objective function @xmath107 in eqn.[eq9 ] by deriving its corresponding gradient on the psd manifold @xmath49 .
* input * : the initial matrix @xmath108 + 1 .
repeat * + 3 .
line search along the geodesic @xmath111 with the direction @xmath112 from @xmath113 to find @xmath114 .
@xmath115 , @xmath116 .
* until * convergence + * output * : the optimized matrix @xmath40 [ alg1 ] as the conjugate gradient algorithm developed in euclidean space , the rcg algorithm on riemannian manifolds is an iterative procedure . as given in algorithm[alg1 ] , an outline for the iterative part of the algorithm goes as follows : at the @xmath73-th iteration , find @xmath117 by searching the minimum of @xmath118 along the geodesic in the direction @xmath119 from @xmath120 , compute the riemannian @xmath121 at this point , choose the new search direction by @xmath122 and iterate until convergence . in the procedure
, the riemannian @xmath121 can be easily approximately from its corresponding euclidean gradient @xmath123 by the computation @xmath124 , and the operation @xmath125 is the parallel transport of tangent vector @xmath119 from @xmath120 to @xmath126 . for more details , we refer readers to @xcite . as for now , we just need to compute the euclidean gradient for our objective function @xmath107 in eqn.[eq9 ] . as the euclidean gradient @xmath127 and its corresponding directional derivatives
are related with the following equality : @xmath128=\langle d_w\mathcal{j}(\bm{w } ) , \dot{\bm{w } } \rangle .
\label{eq10}\ ] ] by employing the basic rule and standard properties of the directional derivatives , @xmath129 $ ] can be derived by : @xmath130 \\ & = \frac{\langle \bm{u}\bm{g}\circ d_w k(\bm{w})[\dot{\bm{w}}]\bm{u } , \bm{g}\circ(\bm{y}\bm{y}^{t } ) \rangle_{\mathcal{f } } \|\mathcal{l}\|_{\mathcal{f}}}{\|\mathcal{l}\|_{\mathcal{f}}^2}\\ - & \frac{\langle \mathcal{l } , \bm{g}\circ(\bm{y}\bm{y}^{t } ) \rangle_{\mathcal{f } } \langle \frac{\mathcal{l}}{\|\mathcal{l}\|_{\mathcal{f } } } , \bm{u}\bm{g}\circ d_w k(\bm{w})[\dot{\bm{w}}]\bm{u } \rangle_{\mathcal{f}}}{\|\mathcal{l}\|_{\mathcal{f}}^2 } \\ & = \langle d_w k(\bm{w})[\dot{\bm{w } } ] , \bm{u } \left ( \frac{\bm{g}\circ(\bm{y}\bm{y}^{t})}{\|\mathcal{l}\|_{\mathcal{f}}}- \frac{\mathcal{j}(\bm{w})\mathcal{l}}{\|\mathcal{l}\|_{\mathcal{f}}^2 } \right ) \bm{u } \rangle_{\mathcal{f } } , \label{eq12 } \end{aligned}\ ] ] where @xmath131 , @xmath77 indicates frobenius inner product , @xmath78 denotes frobenius norm .
accordingly , the key issue in eqn.[eq12 ] is to estimate @xmath132 , where @xmath133 is formulated by eqn.[eq6 ] .
when @xmath71 in eqn.[eq4 ] is the geodesic distance of aim defined in eqn.[eq1 ] , the euclidean gradient of @xmath79 can be derived as : @xmath134 where @xmath135 , @xmath136 . for other affine invariant metrics such as stein divergence @xcite , the corresponding euclidean gradient of @xmath79 with
the geodesic distance function @xmath71 being defined in eqn.[eq1.0 ] can be computed by : @xmath137 where @xmath138 , and hence be able to work in our new proposed framework . when endowing the spd manifold with the non - affine invariant metric lem , it not easy to calculate the euclidean gradient of @xmath132 due to the matrix logarithms in it .
thus , we need to study the problem of the computation of the euclidean gradient for the lem case in the following .
first , we decompose the derivative of lem w.r.t .
@xmath40 into three derivatives with the trace form @xmath139 : @xmath140 * proposition 1 .
* _ the derivatives of the three trace forms @xmath139 in eqn.[eq13 ] can be respectively computed by ( here , @xmath135 , @xmath141 ) _ :
d_w(tr(^2(_i ) ) = 4_i ( _ i)[(_i ) ] .
[ eq14 ] + d_w(tr(^2(_j ) ) = 4_j ( _ j)[(_j ) ] .
[ eq15 ] + & d_w(tr((_i)(_j ) ) + & = 2_i ( _ i)[(_j ) ] + 2_j ( _ j)[(_i ) ] .
[ eq16 ] _ proof .
the three formulas for the gradients with the matrix logarithm correspond to the three ones with rotation matrices in @xcite ( section 5.3 ) , where a detailed proof is given . _ by using * proposition 1 . *
( i.e. eqn.[eq14 ] , eqn.[eq15 ] , eqn.[eq16 ] ) and the sum rule of the directional derivatives , we derive @xmath132 with @xmath71 being the geodesic distance of lem in eqn.[eq4 ] as : @xmath142 \\ & + \bm{b}_j \text{d}\log(\bm{\hat{x}}_j)[\log(\bm{\hat{x}}_j)-\log(\bm{\hat{x}}_i ) ] ) \beta k_{ij}(\bm{w } ) .
\label{eq17 } \end{aligned}\ ] ] to calculate the formula eqn.[eq17 ] , we then apply a function of block triangular matrix developed in @xcite to compute the form of @xmath143 $ ] , which is the directional ( frchet ) derivative of @xmath144 at @xmath145 along @xmath146 .
the following theorem shows that the directional derivative appears as the @xmath147 block of the resulting big matrix when @xmath148 is evaluated at a certain block triangular matrix
. * theorem 1 .
* _ let @xmath148 be @xmath149 times continuously differentiable on @xmath150 and let the spectrum of @xmath151 lie in @xmath150 , where @xmath150 is an open subset of @xmath152 . then _
@xmath153\\ 0 & f(\bm{\hat{x}})\\ \end{bmatrix}. \label{eq18 } \end{aligned}\ ] ] _ proof .
_ the result is proved by najfeld and havel @xcite ( theorem 4.11 ) under the assumption that @xmath7 is analytic . _ by using * theorem 1 * , the directional derivative of the matrix logarithm can be easily computed .
the pseudo matlab code of computing @xmath143 $ ] is simply listed as : n = size(x , 1 ) ; z = zeros(n ) ; a = log([x , h ; z , x ] ) ; d = a(1:n , ( n+1):end ) , where @xmath154 $ ] . with the derived gradient formulas in eqn.[eq11 ] , eqn.[eq19 ] and eqn.[eq17 ] , the euclidean gradient eqn.[eq12 ] of the objective function eqn.[eq9 ] for these cases can be computed to feed into the exploited rcg algorithm working on the psd manifold .
since the global convergence of the rcg algorithm has been well - studied in the survey @xcite , we do not investigate it any further here .
the main time complexity of the algorithm is computing the gradient eqn.[eq12 ] , being @xmath155 ( @xmath156 is the iteration number , @xmath157 is the number of selected samples / pairs , @xmath158 is the dimension of the original / target manifold ) in the lem case . in the experiment
, we will also study the running time of each iteration of the algorithm by varying the number of selected between - class pairs for each spd sample .
in this section , we study the effectiveness of the proposed geometry - aware spd similarity learning ( spdsl ) approach by conducting experimental evaluations for three visual classification tasks including face recognition , material categorization and action recognition . in these three tasks , the spd features have been shown to provide powerful representations for images and videos via set - based covariances@xcite , region covariances @xcite and joint covariance descriptors @xcite .
therefore , they are natural choices to evaluate the proposed spdsl exploiting affine - invariant metric ( aim ) , stein divergence and log - euclidean metric ( lem ) . to evaluate the effectiveness of the proposed spdsl approach , we compare three categories of spd - based learning methods , including basic riemannian metric baseline methods , kernel learning based spd discriminant learning methods and dimensionality reduction based spd discriminant learning methods : 1 .
basic riemannian metrics on spd manifold : + affine - invariant metric ( aim ) @xcite , stein divergence @xcite , log - euclidean metric ( lem ) @xcite 2 .
kernel learning based spd matrix learning methods : + pls - based covariance discriminative learning ( cdl ) @xcite , riemannian sparse representation ( rsr ) @xcite and log - euclidean kernels ( lek ) @xcite 3
. dimensionality reduction based spd matrix learning methods : + log - euclidean metric learning ( leml ) @xcite and spd manifold learning ( spdml - aim and spdml - stein ) @xcite with aim and stein divergence note that , the proposed spdsl belongs to the last category of spd discriminant learning methods . as
this paper focuses on studying the problem of supervised spd discriminant learning , we here only report the performances of the original discriminant learning methods such as spdml rather than those of further coupling them with other classifiers as done in the work @xcite .
in addition , in order to study the discriminant learning power of our proposed framework , we replace its supervised learning scheme with that of spdml but still perform optimization on the exploited solution space .
the adaption of the proposed spdsl is denoted with spdsl - aim@xmath159 , spdsl - stein@xmath159 and spdsl - lem@xmath159 . for rsr ,
the parameter @xmath93 was densely sampled around the order of the mean distance and the parameter @xmath160 is sampled in the range of @xmath161 $ ] . for lek
, there are three implements based on polynomial , exponential and radial basis kernels , which are respectively denoted as lek-@xmath162 , lek-@xmath163 and lek-@xmath164 . for lek-@xmath162 and lek-@xmath163 , we densely sampled the @xmath165 from 1 to 50 . the parameters @xmath93 in lek-@xmath164 and the @xmath160 in the three lek versions were all tuned in the same way as rsr . for leml
, the parameter @xmath166 is tuned in the range of [ 0.1 , 1 , 10 ] , and @xmath167 is tuned from 0.1 to 0.5 .
for spdml and our method spdsl , the maximum iteration number of the optimization algorithm is set to 50 , the parameters @xmath103 is fixed as the minimum number of samples in one class , the dimensionality of the lower - dimensional spd manifold and @xmath106 were tuned by cross - validation .
the parameter @xmath93 in our method is set to @xmath94 , where @xmath95 is equal to the mean distance of all pairs of training data .
[ fig3 ] [ cols="^,^,^,^,^,^,^",options="header " , ] [ tab4 ] we then employ the hdm05 database @xcite to handle with the problem of human action recognition from motion capture sequences . as shown in fig.[fig3 ] , this dataset contains 2,337 sequences of 130 motion classes , e.g. , _ `
clap above head',`lie down floor',`rotate arms ' , ` throw basket ball ' _ , in 10 to 50 realizations executed by various actors .
the 3d locations of 31 joints of the subjects are provided over time acquired at the speed of 120 frames per second . following the previous works @xcite
, we represent an action of a @xmath168 joints skeleton observed over m frames by its joint covariance descriptor .
this descriptor is an form of spd matrix of size @xmath169 , which is computed by the second order statistics of 93-dimensional vectors concatenating the 3d coordinates of the 31 joints in each frame . as the evaluation protocol on uiuc , on this dataset
, we also conduct 10 times random evaluations , in which half of sequences ( around 1,100 sequences ) are randomly selected for training data , and the rest are used for testing . on the hdm05 database ,
the work @xcite only used 14 motion classes for evaluation while we tested these methods for identifying 130 action classes .
table.[tab4 ] summarizes the performances of the comparative algorithms on the uiuc dataset . in the evaluation ,
the dimensions of resulting manifolds achieved by dimensionality reduction methods are all set as 30 .
different from the last two evaluations , cdl and rsr performance worse than other competing methods .
the proposed spdsl again improves the existing dimensionality reduction based methods leml and spdml with 1%-3% , and achieve state - of - the - art performance on the hdm05 database . since our method spdsl and the two methods spdml , leml adopt the same spd matrix learning scheme , we here mainly make two pieces of discussions between them .
first , compared with the related manifold learning method spdml , our spdsl framework proposes a more general solution and a more favorable objective function .
this point has been validated by the three evaluations . as can be seen from table [ tab2 ] , table [ tab3 ] and table [ tab4 ] ,
there are two key conclusions observed from the three visual recognition tasks : \a ) as for the new solution , its main benefits lie in enlarging the search domain and opening up the possibility of using non - affine invariant metrics ( e.g. lem ) . while spdml * for affine invariant metrics aim and stein improves spdml mildly ( this may depend on the data ) , the gains of spdml*-lem over the aim and stein cases are relatively obvious , i.e. 1.65% , 2.15% , 6.21% on average , respectively for the three datasets .
\b ) the new objective function ( for similarity regression ) is quite different from that ( for graph embedding ) used in [ 5 ] . while it s hard to theoretically prove the gains , we have empirically studied its priority . by comparing spdsl with spdml * , the improvements for the three datasets are 2.13% , 1.03% , 6.34% on average for the three used databases , respectively .
second , in contrast to leml which focuses on metric learning , our spdsl learns discriminative similarities on spd manifolds . besides , while leml performs metric learning on the tangent space of spd manifolds , the proposed spdsl learns similarity directly on the spd manifolds .
intuitively , our learning scheme would more faithfully respect the riemannian geometry of the data space , and thus could lead to more favorable spd features for classification tasks . from the above three evaluations , we can see some improvements of spdsl over leml .
we have proposed a geometry - aware spd similarity learning ( spdsl ) framework for more robust visual classification tasks . under this framework , by exploiting the riemannian geometry of psd manifolds , we open the possibility of directly learning the manifold - manifold transformation matrix . to achieve the discriminant learning on the spd features , this work devises a new spdsl technique working on spd manifolds . with the objective of the proposed spdsl
, we derive an optimization algorithm on psd manifolds to pursue the transformation matrix .
extensive evaluations have studied both the effectiveness of efficiency of our spdsl on three challenging datasets . for future work , the study on the relationship between the selected riemannian metrics of psd manifolds and spd manifolds would be interesting for the problem of supervised spd similarity learning . besides , if neglecting the designed discriminant function on spd features , learning the transformation on spd features for object sets is equal to learning the projection on single object features .
thus , this work can be extended to learn hierarchical representations on object feature by leveraging the current powerful deep learning techniques .
this work has been carried out mainly at the institute of computing technology ( ict ) , chinese academy of sciences ( cas ) .
it is partially supported by 973 program under contract no .
2015cb351802 , natural science foundation of china under contracts nos .
61390511 , 61173065 , 61222211 , and 61379083 .
m. t. harandi , c. sanderson , r. hartley , and b. c. lovell , `` sparse coding and dictionary learning for symmetric positive definite matrices : a kernel approach , '' in _ proc . euro
comput . vision _ , 2012 .
m. e. hussein , m. torki , m. a. gowayyed , and m. el - saban , `` human action recognition using a temporal hierarchy of covariance descriptors on 3d joint locations , '' in _ international joint conf . on artificial intelligence _
, 2013 .
v. arsigny , p. fillard , x. pennec , and n. ayache , `` geometric means in a novel vector space structure on symmetric positive - definite matrices , '' _ siam j. matrix analysis and applications _ , vol . 29 , no . 1 ,
pp . 328347 , 2007 .
m. faraki , m. t. harandi , and f. porikli , `` approximate infinite - dimensional region covariance descriptors for image classification , '' in _ international conference on acoustics , speech and signal processing _ , 2015 , pp .
13641368 .
m. t. harandi , r. hartley , b. lovell , and c. sanderson , `` sparse coding on symmetric positive definite manifolds using bregman divergences , '' _ ieee transactions on neural networks and learning systems _ , vol .
27 , no . 6 , pp .
12941306 , 2016 .
z. huang , r. wang , s. shan , x. li , and x. chen .
, `` log - euclidean metric learning on symmetric positive definite manifold with application to image set classification , '' in _ proc .
_ , 2015 .
s. bonnabel and r. sepulchre , `` riemannian metric and geometric mean for positive semidefinite matrices of fixed rank , '' _ siam journal on matrix analysis and applications _ ,
31 , no . 3 , pp .
10551070 , 2009 .
s. yan , d. xu , b. zhang , h. zhang , q. yang , and s. lin , `` graph embedding and extensions : a general framework for dimensionality reduction , '' _ ieee trans . on pattern
_ , vol . 29 , no . 1 ,
pp . 4051 , 2007 .
a. al - mohy and n. higham ,
`` computing the frchet derivative of the matrix exponential , with an application to condition number estimation , '' _ siam journal on matrix analysis and applications _
30 , no . 4 , pp .
16391657 , 2009 . | symmetric positive definite ( spd ) matrices have been widely used for data representation in many visual recognition tasks .
the success mainly attributes to learning discriminative spd matrices with encoding the riemannian geometry of the underlying spd manifold . in this paper
, we propose a geometry - aware spd similarity learning ( spdsl ) framework to learn discriminative spd features by directly pursuing manifold - manifold transformation matrix of column full - rank .
specifically , by exploiting the riemannian geometry of the manifold of fixed - rank positive semidefinite ( psd ) matrices , we present a new solution to reduce optimizing over the space of column full - rank transformation matrices to optimizing on the psd manifold which has a well - established riemannian structure . under this solution , we exploit a new supervised spd similarity learning technique to learn the transformation by regressing the similarities of selected spd data pairs to their ground - truth similarities on the target spd manifold . to optimize the proposed objective function
, we further derive an algorithm on the psd manifold .
evaluations on three visual classification tasks show the advantages of the proposed approach over the existing spd - based discriminant learning methods .
discriminative spd matrices , riemannian geometry , spd manifold , geometry - aware spd similarity learning , psd manifold . |
the stochastic block model is the simplest statistical model for networks with a community ( or cluster ) structure . as such ,
it has attracted considerable amount of work across statistics , machine learning , and theoretical computer science @xcite .
a random graph @xmath1 from this model has its vertex set @xmath2 partitioned into @xmath3 groups , which are assigned @xmath3 distinct labels .
the probability of edge @xmath4 being present depends on the group labels of vertices @xmath5 and @xmath6 . in the context of social network analysis
, groups correspond to social communities @xcite . for other data - mining applications , they represent latent attributes of the nodes @xcite . in all of these cases ,
we are interested in inferring the vertex labels from a single realization of the graph .
in this paper we develop an information - theoretic viewpoint on the stochastic block model .
namely , we develop an explicit ( ` single - letter ' ) expression for the per - vertex conditional entropy of the vertex labels given the graph .
equivalently , we compute the asymptotic per - vertex mutual information between the graph and the vertex labels .
our results hold asymptotically for large networks under suitable conditions on the model parameters .
the asymptotic mutual information is of independent interest , but is also intimately related to estimation - theoretic quantities . for the sake of simplicity
, we will focus on the symmetric two group model .
namely , we assume the vertex set @xmath7\equiv \{1,2,\dots , n\}$ ] to be partitioned into two sets @xmath8 , with @xmath9 independently across vertices @xmath5 .
in particular , the size of each group @xmath10 concentrates tightly around its expectation @xmath11 .
conditional on the edge labels , edges are independent with @xmath12 throughout we will denote by @xmath13 the set of vertex labels @xmath14 , and we will be interested in the conditional entropy @xmath15 or equivalently the mutual information @xmath16 in the limit @xmath17
. we will write @xmath18 ( or @xmath19 ) to imply that the graph @xmath0 is distributed according to the stochastic block model with @xmath20 vertices and parameters @xmath21 .
since we are interested in the large @xmath20 behavior , two preliminary remarks are in order : 1 . _
_ we obviously have , and entropies will be measured in nats . ]
it is therefore natural to study the per - vertex entropy @xmath23 . + as we will see ,
depending on the model parameters , this will take any value between @xmath24 and @xmath25 .
2 . _ scaling . _
the reconstruction problem becomes easier when @xmath26 and @xmath27 are well separated , and more difficult when they are closer to each other .
for instance , in an early contribution , dyer and frieze @xcite proved that the labels can be reconstructed exactly modulo an overall flip if @xmath28 are distinct and independent of @xmath20 .
this in particular implies @xmath29 in this limit ( in fact , it implies @xmath30 ) . in this regime ,
the ` signal ' is so strong that the conditional entropy is trivial .
indeed , recent work @xcite show that this can also happen with @xmath26 and @xmath27 vanishing , and characterizes the sequences @xmath31 for which this happens .
( see section [ sec : related ] for an account of related work . )
+ let @xmath32 be the average edge probability .
it turns out that the relevant ` signal - to - noise ratio ' ( snr ) is given by the following parameter : @xmath33 indeed , we will see that @xmath23 of order @xmath34 , and has a strictly positive limit when @xmath35 is of order one .
this is also the regime in which the fraction of incorrectly labeled vertices has a limit that is strictly between @xmath24 and @xmath34 .
as mentioned above , our main result provides a single - letter characterization for the per - vertex mutual information .
this is given in terms of an _ effective gaussian scalar channel_. namely , define the gaussian channel @xmath36 where @xmath37 independent subscript ] of @xmath38 .
we denote by @xmath39 and @xmath40 the corresponding minimum mean square error and mutual information : @xmath41 in the present case , these quantities can be written explicitly as gaussian integrals of elementary functions : @xmath42 we are now in position to state our main result .
[ thm : main ] for any @xmath43 , let @xmath44 be the largest non - negative solution of the equation : @xmath45 we refer to @xmath46 as to the _ effective signal - to - noise ratio_. further , define @xmath47 by : @xmath48 let the graph @xmath49 and vertex labels @xmath50 be distributed according to the stochastic block model with @xmath20 vertices and parameters @xmath51 ( i.e. @xmath52 ) and define @xmath53
. assume that , as @xmath17 , @xmath54 @xmath55 and @xmath56 @xmath57 .
then , @xmath58 a few remarks are in order . of course
, we could have stated our result in terms of conditional entropy . namely @xmath59
notice that our assumptions require @xmath57 _ at any , arbitrarily slow , rate_. in words , this corresponds to the graph average degree diverging at any , arbitrarily slow , rate .
recently ( see section [ sec : related ] for a discussion of this literature ) , there has been considerable interest in the case of bounded average degree , namely @xmath60 with @xmath61 bounded .
our proof gives an explicit error bound in terms of problem parameters even when @xmath62 is of order one .
hence we are able to characterize the asymptotic mutual information for large - but - bounded average degree up to an offset that vanishes with the average degree .
explicitly , we prove that : @xmath63 for some absolute constant @xmath64 . our main result and its proof has implications on the minimum error that can be achieved in estimating the labels @xmath50 from the graph @xmath49 . for reasons that will become clear below , a natural metric is given by the matrix minimum mean square error @xmath65 ( occasionally , we will also use the notation @xmath66 for @xmath67 . ) using the exchangeability of the indices @xmath68 , this can also be rewritten as @xmath69 ^ 2\big\}\\ & = \e\big\{\big[x_1x_2-\e\{x_1x_2|\bg\}\big]^2\big\}\label{eq : mmse_2vert}\\ & = \min_{\hx_{12 } : \cg_n\to \reals } \e\big\{\big[x_1x_2-\hx_{12}(\bg)\big]^2\big\}\ , .
\label{eq : mmse_2vert_opt } \ ] ] ( here @xmath70 denotes the set of graphs with vertex set @xmath71 $ ] . ) in words , @xmath67 is the minimum error incurred in estimating the relative sign of the labels of two given ( distinct ) vertices .
equivalently , we can assume that vertex @xmath34 has label @xmath72 .
then @xmath67 is the minimum mean square error incurred in estimating the label of any other vertex , say vertex @xmath73 .
namely , by symmetry , we have ( see section [ sec : estimation ] ) @xmath74 ^ 2\big|x_1=+1\big\}\label{eq : alternativemmse1}\\ & = \min_{\hx_{2|1 } : \cg_n\to\reals } \e\big\{\big[x_2-\hx_{2|1}(\bg)\big]^2|x_1=+1\big\}\ , .
\label{eq : alternativemmse2 } \ ] ] in particular @xmath75 $ ] , with @xmath76 corresponding to random guessing .
[ thm : mainestimation ] under the assumptions of theorem [ thm : main ] ( in particular assuming @xmath77 as @xmath17 ) , the following limit holds for the matrix minimum mean square error @xmath78 further , this implies @xmath79 for @xmath80 and @xmath81 for @xmath82 . for further discussion of this result and its generalizations
, we refer to section [ sec : estimation ] .
in particular , corollary [ coro : metrics ] establishes that @xmath83 is a phase transition for other estimation metrics as well , in particular for overlap and vector mean square error . as theorem [ thm : main ] ,
also the last theorem holds under the mild condition that the average degree @xmath84 diverges _ at any , arbitrarily slow rate_. this should be contrasted with the phase transition of naive spectral methods .
it is well understood that the community structure can be estimated by the principal eigenvector of the centered adjacency matrix @xmath85 .
( we denote by @xmath49 the graph as well as its adjacency matrix . )
this approach is successful fro @xmath82 but requires average degree @xmath86 for @xmath87 a constant @xcite .
our proof of theorem [ thm : main ] and theorem [ thm : mainestimation ] involves the analysis of a gaussian observation model , whereby the rank one matrix @xmath88 is corrupted by additive gaussian noise , according to @xmath89 .
in particular , we prove a single letter characterization of the asymptotic mutual information per dimension in this model @xmath90 , cf .
theorem [ thm : singlelettergauss ] below .
the resulting asymptotic value is proved to coincide with the asymptotic value in the stochastic block model , as established in theorem [ thm : main ] .
in other words , the per - dimension mutual information turns out to be _
universal _ across multiple noise models . in section [ sec : related ] we review the literature on this problem .
we then discuss the connection with estimation in section [ sec : estimation ] .
this section also demonstrates how to evaluate the asymptotic formula in theorem [ thm : main ] .
section [ sec : strategy ] describes the proof strategy . as an intermediate step ,
we introduce a gaussian observation model which is of independent interest .
the proof of theorem [ thm : main ] is reduced to two main propositions : * proposition [ prop : gausseq ] establishes that within the regime defined in theorem [ thm : main] the stochastic block model is asymptotically equivalent to the gaussian observation model ( see section [ sec : strategy ] for a formal definition ) .
this statement ( with explicit error bounds ) is proved in section [ sec : gausseq ] through a careful application of the lindeberg method .
* proposition [ prop : singleletter ] develops a single - letter characterization of the asymptotic per - vertex mutual information of the gaussian observation model .
the proof of this fact is presented in section [ sec : singleletterproof ] and builds on two steps .
we first prove an asymptotic upper bound on the matrix minimum mean square error @xmath67 using an approximate message passing ( amp ) algorithm .
we then use an area theorem to prove that this upper bound is tight .
finally , section [ sec : proofestimation ] contains the proof of theorem [ thm : mainestimation ] .
several technical details are deferred to the appendices .
the set of first @xmath20 integers is denoted by @xmath71 = \{1,2,\dots , n\}$ ] .
when possible , we will follow the convention of denoting random variables by upper - case letters ( e.g. @xmath91 ) , and their values by lower case letters ( e.g. @xmath92 ) .
we use boldface for vectors and matrices , e.g. @xmath50 for a random vector and @xmath93 for a deterministic vector .
the graph @xmath49 will be identified with its adjacency matrix .
namely , with a slight abuse of notation , we will use @xmath49 both to denote a graph @xmath94 , e)$ ] ( with @xmath7 $ ] the vertex set , and @xmath95 the edge set , i.e. a set of unordered pairs of vertices ) , and its adjacency matrix .
this is a symmetric zero - one matrix @xmath96 with entries @xmath97 throughout we assume @xmath98 by convention .
we write @xmath99 to mean that @xmath100 for a universal constant @xmath64 .
we denote by @xmath64 a generic ( large ) constant that is independent of problem parameters , whose value can change from line to line .
we say that an event holds _ with high probability _ if it holds with probability converging to one as @xmath17 .
we denote the @xmath101 norm of a vector @xmath93 by @xmath102 and the frobenius norm of a matrix @xmath103 by @xmath104 .
the ordinary scalar product of vectors @xmath105 is denoted as @xmath106 .
unless stated otherwise , logarithms will be taken in the natural basis , and entropies measured in nats .
the stochastic block model was first introduced within the social science literature in @xcite . around the same time , it was studied within theoretical computer science @xcite , under the name of ` planted partition model . '
a large part of the literature has focused on the problem of _ exact recovery _ of the community ( cluster ) structure .
a long series of papers @xcite , establishes sufficient conditions on the gap between @xmath26 and @xmath27 that guarantee exact recovery of the vertex labels with high probability .
a sharp threshold for exact recovery was obtained in @xcite , showing that for @xmath107 , @xmath108 , @xmath109 , exact recovery is solvable ( and efficiently so ) if and only if @xmath110 .
efficient algorithms for this problem were also developed in @xcite .
for the sbm with arbitrarily many communities , necessary and sufficient conditions for exact recovery were recently obtained in @xcite .
the resulting sharp threshold is efficiently achievable and is stated in terms of a ch - divergence .
a parallel line of work studied the _ detection _ problem . in this case , the estimated community structure is only required to be asymptotically positively correlated with the ground truth . for this requirement , two independent groups @xcite proved that detection is solvable ( and so efficiently ) if and only if @xmath111 , when @xmath112 , @xmath113 .
this settles a conjecture made in @xcite and improves on earlier work @xcite .
results for detection with more than two communities were recently obtained in @xcite .
a variant of community detection with a single hidden community in a sparse graph was studied in @xcite . in a sense , the present paper bridges detection and exact recovery , by characterizing the minimum estimation error when this is non - zero , but for @xmath82 smaller than for random guessing .
an information - theoretic view of the sbm was first introduced in @xcite .
there it was shown that in the regime of @xmath112 , @xmath113 , and @xmath114 ( i.e. , disassortative communities ) , the normalized mutual information @xmath115 admits a limit as @xmath17 .
this result is obtained by showing that the condition entropy @xmath15 is sub - additive in @xmath20 , using an interpolation method for planted models . while the result of @xcite holds for arbitrary @xmath116 ( possibly small ) and extend to a broad family of planted models ,
the existence of the limit in the assortative case @xmath117 is left open .
further , sub - additivity methods do not provide any insight as to the limit value . for the partial recovery of the communities
, it was shown in @xcite that the communities can be recovered up to a vanishing fraction of the nodes if and only if @xmath118 diverges .
this is generalized in @xcite to the case of more than two communities . in these regimes
, the normalized mutual information @xmath115 ( as studied in this paper ) tends to @xmath25 nats . for the constant degree regime , it was shown in @xcite that when @xmath119 is sufficiently large , the fraction of nodes that can be recovered is determined by the broadcasting problem on tree @xcite .
namely , consider the reconstruction problem whereby a bit is broadcast on a galton - watson tree with poisson(@xmath120 ) offspring and with binary symmetric channels of bias @xmath121 on each branch .
then the probability of recovering the bit correctly from the leaves at large depth gives the fraction of nodes that can be correctly labeled in the sbm . in terms of proof techniques , our arguments are closest to @xcite .
we use the well - known lindeberg strategy to reduce computation of mutual information in the sbm to mutual information of the gaussian observation model .
we then compute the latter mutual information by developing sharp algorithmic upper bounds , which are then shown to be asymptotically tight via an area theorem .
the lindeberg strategy builds from @xcite while the area theorem argument also appeared in @xcite .
we expect these techniques to be more broadly applicable to compute quantities like normalized mutual information or conditional entropy in a variety of models .
let us finally mentioned that the result obtained in this paper are likely to extend to more general sbms , with multiple communities , to the censored block model studied in @xcite , the labeled block model @xcite , and other variants of block models . in particular , it would be interesting to understand which estimation - theoretic quantities appear for these models , and whether a general result stands behind the case of this paper .
while this paper was in preparation , lesieur , krzakala and zdborov @xcite studied estimation of low - rank matrices observed through noisy memoryless channels .
they conjectured that the resulting minimal estimation error is universal across a variety of channel models .
our proof ( see section [ sec : strategy ] below ) establishes universality across two such models : the gaussian and the binary output channels .
we expect that similar techniques can be useful to prove universality for other models as well .
in this section we discuss how to evaluate the asymptotic formulae in theorem [ thm : main ] and theorem [ thm : mainestimation ] .
we then discuss the consequences of our results for various estimation metrics . before passing to these topics
, we will derive a simple upper bound on the per - vertex mutual information , which will be a useful comparison for our results .
it is instructive to start with an elementary upper bound on @xmath16 .
[ lemma : elementary ] assume @xmath26 , @xmath27 satisfy the assumptions of theorem [ thm : main ] ( in particular @xmath54 @xmath55 and @xmath56 @xmath122 ) .
then @xmath123 we have @xmath124 where @xmath125 follows since @xmath126 are conditionally independent given @xmath50 and @xmath127 because @xmath128 only depends on @xmath50 through the product @xmath129 ( notice that there is no comma but product in @xmath130 . from our model , it is easy to check that @xmath131 the claim follows by substituting @xmath132 , @xmath133 and by taylor expansion for all @xmath20 large enough . ] . ) .
the ` effective signal - to - noise ratio ' @xmath46 is given by the intersection of the curve @xmath134 , and the line @xmath135.,title="fig : " ] ( -90,120)@xmath136 . the dashed lines are simple upper bounds : @xmath137 ( cf
. lemma [ lemma : elementary ] ) and @xmath138 .
right frame : asymptotic estimation error under different metrics ( see section [ sec : est ] ) . note the phase transition at @xmath83 in both frames.,title="fig : " ] .
the dashed lines are simple upper bounds : @xmath137 ( cf .
lemma [ lemma : elementary ] ) and @xmath138 .
right frame : asymptotic estimation error under different metrics ( see section [ sec : est ] ) . note the phase transition at @xmath83 in both frames.,title="fig : " ] our asymptotic expression for the mutual information , cf .
theorem [ thm : main ] , and for the estimation error , cf .
theorem [ thm : mainestimation ] , depends on the solution of eq .
( [ eq : mainequation ] ) which we copy here for the reader s convenience : @xmath139 here we defined @xmath140 the effective signal - to - noise ratio @xmath46 that enters theorem [ thm : main ] and theorem [ thm : mainestimation ] is the largest non - negative solution of eq .
( [ eq : mainequation ] ) .
this equation is illustrated in figure [ fig : formulaevaluation ] .
it is immediate to show from the definition ( [ eq : gdef ] ) that @xmath141 is continuous on @xmath142 with @xmath143 , and @xmath144 .
this in particular implies that @xmath145 is always a solution of eq .
( [ eq : mainequation ] ) .
further , since @xmath39 is monotone decreasing in the signal - to - noise ratio @xmath146 , @xmath147 is monotone increasing . as shown in the proof of remark [ rem : unique ] ( see appendix [ sec : proofunique ] ) , @xmath141 is also strictly concave on @xmath142 .
this implies that eq .
( [ eq : mainequation ] ) as at most one solution in @xmath148 , and a strictly positive solution only exists if @xmath149 .
we summarize these remarks below , and refer to figure [ fig : informationestimation ] for an illustration .
[ lemma : fixedpoint ] the effective snr , and the asymptotic expression for the per - vertex mutual information in theorem [ thm : main ] have the following properties : * for @xmath80 , we have @xmath150 and @xmath151 . * for @xmath80
, we have @xmath152 strictly with @xmath153 as @xmath154 . +
further , @xmath155 strictly with @xmath156 as @xmath157 .
all of the claims follow immediately form the previous remarks , and simple calculus , except the claim @xmath158 for @xmath82 .
this is direct consequence of the variational characterization established below .
we next give an alternative ( variational ) characterization of the asymptotic formula which is useful for proving bounds .
under the assumptions and definitions of theorem [ thm : main ] , we have @xmath159 the function @xmath160 is differentiable on @xmath142 with @xmath161 as @xmath157 .
hence , the @xmath162 is achieved at a point where the first derivative vanishes ( or , eventually , at @xmath24 ) . using the i - mmse relation @xcite , we get @xmath163 hence the minimizer is a solution of eq .
( [ eq : mainequation ] ) .
as shown above , for @xmath80 , the only solution is @xmath150 , which therefore yields @xmath164 as claimed .
for @xmath82 , eq . ( [ eq : mainequation ] )
admits the two solutions : @xmath24 and @xmath165 .
however , by expanding eq .
( [ eq : derpsi ] ) for small @xmath146 , we obtain @xmath166 and hence @xmath167 is a local maximum , which implies the claim for @xmath82 as well .
we conclude by noting that eq .
( [ eq : mainequation ] ) can be solved numerically rather efficiently .
the simplest method consists is by iteration .
namely , we initialize @xmath168 and then iterate @xmath169 . this approach was used for figure [ fig : informationestimation ] .
theorem [ thm : mainestimation ] establishes that a phase transition takes place at @xmath83 for the matrix minimum mean square error @xmath170 defined in eq .
( [ eq : mmsedef ] ) . throughout this section
, we will omit the subscript @xmath20 to denote the @xmath17 limit ( for instance , we write @xmath171 ) .
figure [ fig : informationestimation ] reports the asymptotic prediction for @xmath172 stated in theorem [ thm : mainestimation ] , and evaluated as discussed above .
the error decreases rapidly to @xmath24 for @xmath82 . in this section
we discuss two other estimation metrics . in both cases
we define these metrics by optimizing a suitable risk over a class of estimators : it is understood that randomized estimators are admitted as well . *
the first metric is the _ vector minimum mean square error _ :
@xmath173 note the minimization over the sign @xmath174 : this is necessary because the vertex labels can be estimated only up to an overall flip .
of course @xmath175 $ ] , since it is always possible to achieve vector mean square error equal to one by returning @xmath176 . *
the second metric is the _ overlap _ : @xmath177 again @xmath178 $ ] ( but now large overlap corresponds to good estimation ) . indeed by returning @xmath179 uniformly at random , we obtain @xmath180 .
+ note that the main difference between overlap and vector minimum mean square error is that in the latter case we consider estimators @xmath181 taking arbitrary real values , while in the former we assume estimators @xmath182 taking binary values . in order to clarify the relation between various metrics ,
we begin by proving the alternative characterization of the matrix minimum mean square error in eqs .
( [ eq : alternativemmse1 ] ) , ( [ eq : alternativemmse2 ] ) . letting @xmath67 be defined as per eq . ( [ eq : mmsedef ] ) , we have @xmath183 ^ 2\big|x_1=+1\big\ } \label{eq : alternativemmse1_bis}\\ & = \min_{\hx_{2|1 } : \cg_n\to\reals } \e\big\{\big[x_2-\hx_{2|1}(\bg)\big]^2|x_1=+1\big\}\ , .
\label{eq : alternativemmse2_bis } \ ] ] first note that eq .
( [ eq : alternativemmse2_bis ] ) follows immediately from eq .
( [ eq : alternativemmse1_bis ] ) since conditional expectation minimizes the mean square error ( the conditioning only changes the prior on @xmath50 ) .
in order to prove eq .
( [ eq : alternativemmse1_bis ] ) , we start from eq .
( [ eq : mmse_2vert ] ) .
since the prior distribution on @xmath184 is uniform , we have @xmath185 where in the second line we used the fact that , conditional to @xmath49 , @xmath50 is distributed as @xmath186 .
continuing from eq .
( [ eq : mmse_2vert ] ) , we get @xmath187 ^ 2\big\}\\ & = \frac{1}{2}\ , \e\big\{\big[x_1x_2-\e\{x_2|x_1=+1,\bg\}\big]^2\big|x_1 = + 1\big\}\nonumber\\ & \phantom{aaaa}+ \frac{1}{2}\ , \e\big\{\big[x_1x_2-\e\{x_2|x_1=+1,\bg\}\big]^2\big|x_1 = -1\big\}\\ & = \e\big\{\big[x_2-\e\{x_2|x_1=+1,\bg\}\big]^2\big|x_1 = + 1\big\}\ , , \ ]
] which proves the claim .
the next lemma clarifies the relationship between matrix and vector minimum mean square error .
its proof is deferred to appendix [ sec : vectorvsmatrix ] .
[ lemma : vectorvsmatrix ] with the above definitions , we have @xmath188 finally , a lemma that relates overlap and vector minimum mean square error , whose proof can be found in appendix [ sec : overlapvsvector ] .
[ lemma : overlapvsvector ] with the above definitions , we have @xmath189 as an immediate corollary of these lemmas ( together with theorem [ thm : mainestimation ] and lemma [ lemma : fixedpoint ] ) , we obtain that @xmath83 is the critical point for other estimation metrics as well .
[ coro : metrics ] the vector minimum mean square error and the overlap exhibit a _ phase transition at @xmath83_. namely , under the assumptions of theorem [ thm : main ] ( in particular , @xmath55 and @xmath57 ) , we have * if @xmath80 , then estimation can not be performed asymptotically better than without any information : @xmath190 * if @xmath82 , then estimation can be performed better than without any information , even in the limit @xmath17 : @xmath191
in this section we describe the main elements used in the proof of theorem [ thm : main ] : * we describe a gaussian observation model which has asymptotically the same mutual information as the sbm introduced above . *
we state an asymptotic characterization of the mutual information of this gaussian model . *
we describe an approximate message passing ( amp ) estimation algorithm that plays a key role in the last characterization .
we then use these technical results ( proved in later sections ) to prove theorem [ thm : main ] in section [ sec : prooftheoremmain ] .
we recall that @xmath192 .
define the gap @xmath193 .
we will assume for the proofs that @xmath194 ( i.e. the assortative model ) but the results also hold for @xmath195 in an analogous fashion .
the edges @xmath126 are conditionally independent given the vertex labels @xmath50 , with distribution : @xmath196 as a first step , we compare the sbm with an alternate gaussian observation model defined as follows .
let @xmath197 be a gaussian random symmetric matrix generated with independent entries @xmath198 and @xmath199 , independent of @xmath50 .
consider the noisy observations @xmath200 defined by @xmath201 note that this model matches the first two moments of the original model .
more precisely , if we define the rescaled adjacency matrix @xmath202 , then @xmath203 and @xmath204 .
our first proposition proves that the mutual information between the vertex labels @xmath50 and the observations agrees to leading order across the two models . [ prop : gausseq ] assume that , as @xmath17 , @xmath54 @xmath55 and @xmath56 @xmath57 .
then there is a constant @xmath64 independent of @xmath20 such that @xmath205 the proof of this result is presented in section [ sec : gausseq ] .
the next step consists in analyzing the gaussian model ( [ eq : gaussobs ] ) , which is of independent interest .
it turns out to be convenient to embed this in a more general model whereby , in addition to the observations @xmath103 , we are also given observations of @xmath50 through a binary erasure channel with erasure probability @xmath206 , @xmath207 .
we will denote by @xmath208 the output of this channel , where we set @xmath209 every time the symbol is erased .
formally we have @xmath210 where @xmath211 are independent random variables , independent of @xmath50 , @xmath49 . in the special case @xmath212 , all of these observations are trivial , and we recover the original model .
the reason for introducing the additional observations @xmath213 is the following .
the graph @xmath49 has the same distribution conditional on @xmath50 or @xmath186 , hence it is impossible to recover the sign of @xmath50 . as we will see , the extra observations @xmath213 allow to break this trivial symmetry and we will recover the required results by continuity in @xmath214 as the extra information vanishes . indeed , our next result establishes a single letter characterization of @xmath215 in terms of a recalibrated _ scalar _ observation problem .
namely , we define the following observation model for @xmath216 a rademacher random variable : @xmath217 here @xmath218 , @xmath219 , @xmath220 , are mutually independent .
we denote by @xmath221 , the minimum mean squared error of estimating @xmath218 from @xmath222 , @xmath223 , conditional on @xmath224 .
recall the definitions ( [ eq : infodef ] ) , ( [ eq : mmsedef ] ) of @xmath40 , @xmath39 , and the expressions ( [ eq : infoformula ] ) , ( [ eq : mmseformula ] ) .
a simple calculation yields @xmath225 [ prop : singleletter ] for any @xmath43 , @xmath226 $ ] , let @xmath227 be the largest non - negative solution of the equation : @xmath228 further , define @xmath229 by : @xmath230 then , we have @xmath231 using continuity in @xmath214 , the last result implies directly a limit result for the mutual information under the gaussian model , which we single out since it is of independent interest .
[ thm : singlelettergauss ] for any @xmath43 , let @xmath46 be the largest non - negative solution of the equation : @xmath232 further , define @xmath47 by : @xmath233 then , we have @xmath234 to analyze the gaussian model eq .
we introduce an approximate message passing ( amp ) algorithm that computes estimates @xmath235 at time @xmath236 , which are functions of the observations @xmath237 .
this construction follows the general scheme of amp algorithms developed in @xcite .
given a sequence of functions @xmath238 , we set @xmath239 and compute @xmath240 above ( and in the sequel ) we extend the function @xmath241 to vectors by applying it component - wise , i.e. @xmath242 . the amp iteration above proceeds analogously to the usual power iteration to compute principal eigenvectors , but has an additional memory term @xmath243 .
this additional term changes the behavior of the iterates in an important way : unlike the usual power iteration , there is an explicit distributional characterization of the iterates @xmath244 in the limit of large dimension .
namely , for each time @xmath236 we will show that , approximately @xmath245 is a scaled version of the truth @xmath246 observed through gaussian noise of a certain variance .
we define the following two - parameters recursion , with initialization @xmath247 , which will be referred to as _ state evolution _ :
@xmath248 where expectation is with respect to the independent random variables @xmath249 , @xmath38 and @xmath250 , setting @xmath251 .
the following lemma makes this distributional characterization precise .
it follows from the more general result of @xcite and we provide a proof in appendix [ app : addproofs ] . [ lem : stateevollem ] let @xmath252 be a sequence of functions such that @xmath253 are lipschitz continuous in their first argument ( where @xmath254 denotes the derivative of @xmath241 with respect to the first argument ) .
let @xmath255 be a test function such that @xmath256 for all @xmath257 .
then the following limit holds almost surely for @xmath258 random variables distributed as above @xmath259 although the above holds for a relatively broad class of functions @xmath241 , we are interested in the amp algorithm for specific functions @xmath241 . specifically , we following sequence of functions @xmath260 it is easy to see that @xmath241 satisfy the requirement of lemma [ lem : stateevollem ]
. we will refer to this version of amp as _ bayes - optimal amp_. note that the definition ( [ eq : optfunc ] ) depends itself on @xmath261 and @xmath262 defined through eqs .
( [ eq : stateevol1 ] ) , ( [ eq : stateevol1 ] ) .
this recursive definition is perfectly well defined and yields @xmath263 using the fact that @xmath264 is the minimum mean square error estimator , we obtain @xmath265 where @xmath266 is given explicitly by eq . ( [ eq : mmseformula ] ) . in other words
, the state evolution recursion reduces to a simple one - dimensional recursion that we can write in terms of the variable @xmath267 .
we obtain @xmath268 our proof strategy uses the amp algorithm to construct estimates that _ bound from above _ the minimum error of estimating @xmath50 from observations @xmath269 .
however , in the limit of a large number of iterations , we show that the gap between this upper bound and the minimum estimation error vanishes via an area theorem .
more explicitly , we develop an upper bound on the matrix mean square error first introduced in eq .
( [ eq : mmsedef ] ) .
we generalize this in the obvious way to the gaussian observation model : @xmath270 ( note that we adopt here a slightly different normalization with respect to eq .
( [ eq : mmsedef ] ) .
this change is immaterial in the large @xmath20 limit . )
we then use amp to construct the sequence of estimators @xmath271 , indexed by @xmath272 , where @xmath273 is defined as in eq .. the matrix mean squared error of this estimators will be denoted by @xmath274 we also define the limits @xmath275 in the course of the proof , we will also see that these limits are well - defined , using the state evolution lemma [ lem : stateevollem ] the proof is almost immediate given propositions [ prop : gausseq ] and [ prop : singleletter ] .
firstly , note that , for any @xmath226 $ ] , @xmath276 since , by proposition [ prop : singleletter ] @xmath277 has a well - defined limit as @xmath17 , and @xmath278 is arbitrary , we have that : @xmath279 it is immediate to check that @xmath229 is continuous in @xmath280 , @xmath281 and @xmath282 as defined in theorem [ thm : main ] . furthermore , as @xmath283 , the unique positive solution @xmath227 of eq .
( [ eq : epsfixedpoint ] ) converges to @xmath46 , the largest non - negative solution to of eq .
( [ eq : mainequation ] ) , which we copy here for the readers convenience : @xmath284 this follows from the smoothness and concavity of the function @xmath136 ( see lemma [ rem : unique ] ) .
it follows that @xmath285 and therefore @xmath286 this proves theorem [ thm : singlelettergauss ] .
theorem [ thm : main ] follows by applying proposition [ prop : gausseq ] .
given a collection @xmath287 of random variables defined on the same probability space as @xmath50 , and a non - negative real number @xmath288 , we define the following hamiltonian and log - partition function associated with it : @xmath289 [ lem : gaussentro ] we have the identity : @xmath290 by definition : @xmath291 since the two distributions @xmath292 and @xmath293 are absolutely continuous with respect to each other , we can write the above simply in terms of the ratio of ( lebesgue ) densities , and we obtain : @xmath294 we modify the final term as follows : @xmath295 substituting this in eq .
we have @xmath296 as required .
[ lem : sbmentro ] define the ( random ) hamiltonian @xmath297 by : @xmath298 then we have that : @xmath299 this follows directly from the definition of mutual information : @xmath300 as in lemma [ lem : gaussentro ] we can write this in terms of densities as : @xmath301 substituting this in the mutual information formula eq .
yields the lemma .
define the random variables @xmath302 as follows : @xmath303 the following lemma shows that , to compute @xmath16 it suffices to compute the log - partition function with respect to the approximating hamiltonian .
assume that , as @xmath17 , @xmath54 @xmath55 and @xmath56 @xmath57 .
then , we have @xmath304 [ lem : sbmapproxrel ] we concentrate on the log - partition function for the hamiltonian @xmath305 . first , using the fact that @xmath306 when @xmath307 : @xmath308 now when @xmath309 , for small enough @xmath310 , we have by taylor expansion the following approximation for @xmath311 $ ] : @xmath312 which implies , by triangle inequality : @xmath313 where @xmath314 we first simplify the rhs in eq .. recalling the definition of @xmath315 : @xmath316 this implies that : @xmath317 where @xmath318 satisfies eq .. we now use the following remark , which is a simple application of bernstein inequality ( the proof is deferred to appendix [ app : addproofs ] ) . [ rem : berns ] there exists a constant @xmath64 such that for every @xmath20 large enough : @xmath319 using this remark , the error bound eq . and eq . in lemma
[ lem : sbmentro ] yields @xmath320 substituting @xmath321 gives the lemma .
we now control the deviations that occur when replacing the variables @xmath315 with gaussian variables @xmath322 .
[ lem : approx ] assume that , as @xmath17 , @xmath54 @xmath55 and @xmath56 @xmath57 . then we have : @xmath323 this proof follows the lindeberg strategy @xcite
. we will show that : @xmath324 ( with the @xmath325 term uniform in @xmath50 ) .
the claim then follows by taking expectations on both sides .
note that , by construction : @xmath326 and @xmath327 we now derive the following estimates : @xmath328 here @xmath329 is the @xmath3-fold derivative of @xmath330 in the entry @xmath331 of the matrix @xmath332 . to write explicitly the derivatives @xmath333
we introduce some notation . for a function @xmath334
, we write @xmath335 to denote its expectation with respect to the measure defined by the hamiltonian @xmath336 . explicitly : @xmath337 then the partial derivatives above can be expressed as @xmath338 however since @xmath339 , we obtain : @xmath340 applying theorem 2 of @xcite ( stated below as theorem [ thm : lindegen ] ) gives : @xmath341 further , we have : @xmath342 here @xmath343 denotes the derivative with respect to the variable @xmath288 .
thus , @xmath344 combining eqs . , gives eq .
, and the lemma follows by taking expectations on either side .
we state below the lindeberg generalization theorem for convenience : [ thm : lindegen ] suppose we are given two collections of random variables @xmath345}$ ] , @xmath346}$ ] with independent components and a function @xmath347 .
let @xmath348 and @xmath349 .
then : @xmath350 with these in hand , we can now prove proposition [ prop : gausseq ] .
the proposition follows simply by combining the formulae for @xmath351 in lemmas [ lem : gaussentro ] , [ lem : sbmapproxrel ] with the approximation guarantee of lemma [ lem : approx ] . throughout this section
we will write @xmath352 whenever we want to emphasize the dependence of the law of @xmath103 on the signal to noise parameter @xmath288 .
the proof of proposition [ prop : singleletter ] follows essentially from a few preliminary lemmas .
we begin with some properties of the fixed point equation ( [ eq : epsfixedpoint ] ) .
the proof of this lemma can be found in appendix [ sec : proofunique ] .
[ rem : unique ] for any @xmath353 $ ] , the following properties hold for the function @xmath354 : 1 .
it is continuous , monotone increasing and concave in @xmath355 .
it satisfies the following limit behaviors @xmath356 & = 1\ , .
\end{aligned}\ ] ] as a consequence we have the following for all @xmath357 $ ] : 1 . a non - negative solution @xmath358 of eq .
( [ eq : epsfixedpoint ] ) exists and is unique for all @xmath278 .
2 . for any @xmath278
, the function @xmath359 is differentiable in @xmath288 .
3 . let @xmath360 be defined recursively by eq .
( [ eq : simplstateevol3 ] ) , with initialization @xmath361 . then @xmath362 .
we then compute the value of @xmath363 at @xmath364 and @xmath365 .
[ lem : limpoints ] for any @xmath278 : @xmath366 recall the definition of @xmath39 , cf .
( [ eq : mmsedef ] ) .
upper bounding @xmath39 by the minimum error obtained by linear estimator yields , for any @xmath367 , @xmath368 .
substituting these bounds in eq .
( [ eq : epsfixedpoint ] ) , we obtain @xmath369 = \lambda-(1-\eps)+o(\lambda^{-1})\ , , \ ] ] where the last expansion holds as @xmath154 .
let us now consider the limit @xmath370 , cf .
claim ( [ eq : psilamblda0 ] ) . considering eq .
( [ eq : psieps ] ) , and using @xmath371 , we have @xmath372 .
further from the definition ( [ eq : infodef ] ) it follows ) . ]
that @xmath373 thus yielding eq .
( [ eq : psilamblda0 ] ) .
consider next the @xmath154 limit of eq .
( [ eq : psilambldainfty ] ) . in this limit
( [ eq : boundgamma ] ) implies @xmath374 , where @xmath375 .
hence @xmath376 ( this follows again from the definition of @xmath40 ) .
further @xmath377 substituting in eq .
( [ eq : psieps ] ) we obtain the desired claim . the next lemma characterizes the limiting matrix mean squared error of the amp estimates .
[ lem : fixedpt ] let @xmath360 be defined recursively by eq .
( [ eq : simplstateevol3 ] ) with initialization @xmath378 , and recall that @xmath358 denotes the unique non - negative solution of eq .. then the following limits hold for the amp mean square error @xmath379 note that eq .
( [ eq : mseamp_tinfty ] ) follows from eq .
( [ eq : mseamp_t ] ) using lemma [ rem : unique ] , point @xmath380 .
we will therefore focus on proving eq .
( [ eq : mseamp_t ] ) .
first notice that : @xmath381 since @xmath382 , the first term evaluates to 1 .
we use lemma [ lem : stateevollem ] to deal with the final two terms . consider the last term @xmath383 . using lemma [ lem : stateevollem ] with the @xmath384 we have ,
almost surely @xmath385 note also that @xmath386 and @xmath387 are bounded by 1 , hence so is @xmath388 .
it follows from the bounded convergence theorem that @xmath389 in a similar manner , we have that @xmath390 , whence the thesis follows .
[ lem : psiintegral ] for every @xmath391 and @xmath392 : @xmath393 by differentiating eq .
( [ eq : psieps ] ) we obtain ( recall @xmath394 ) : @xmath395 it follows from the uniqueness and differentiability of @xmath358 ( cf .
lemma [ rem : unique ] ) that @xmath396 is differentiable for any fixed @xmath278 , with derivative @xmath397 the lemma follows from the fundamental theorem of calculus using lemma [ lem : limpoints ] for @xmath398 , and lemma [ lem : fixedpt ] , cf .
( [ eq : mseamp_tinfty ] ) .
we are now in a position to prove proposition [ prop : singleletter ] .
we start from a simple remark , proved in appendix [ rem : limitentro ] we have @xmath399 further the asymptotic mutual information satisfies @xmath400 we defer the proof of these facts to appendix [ sec : limentro ] . applying the ( conditional )
i - mmse identity of @xcite we have @xmath401 we therefore have @xmath402 \\ & \stackrel{(b)}{= } \lim_{\lambda\to\infty}\liminf_{n\to\infty } \frac{1}{4 } \int_{0}^{\lambda}\mmse(\lambda',\eps , n ) \ , \d\lambda'\\ & \stackrel{(c)}{\le } \lim_{\lambda\to\infty}\lim_{t\to \infty}\limsup_{n\to\infty } \frac{1}{4 } \int_{0}^{\lambda}\mseamp(t;\lambda',\eps , n ) \
, \d\lambda'\\ & \stackrel{(d)}{= } \lim_{\lambda\to\infty}\lim_{t\to \infty } \frac{1}{4 } \int_{0}^{\lambda}\mseamp(t;\lambda',\eps ) \ , \d\lambda ' \ ] ] where @xmath125 follows from remark [ rem : limitentro ] , @xmath127 from eq .
( [ eq : ixx ] ) and ( [ eq : immse ] ) , @xmath403 from ( [ eq : immse2 ] ) , and @xmath380 from bounded convergence . continuing from the previous chain
we get @xmath404\\ & \stackrel{(g)}{= } ( 1-\eps)\log 2\ , , \ ] ] where @xmath405 follows from lemma [ lem : fixedpt ] , @xmath406 from lemma [ lem : psiintegral ] , and @xmath407 from lemma [ lem : limpoints ] .
we therefore have a chain of equalities , whence the inequality @xmath403 must hold with equality . since @xmath408 for any @xmath288 , this implies @xmath409 for almost every @xmath288 .
the conclusion follows for every @xmath288 by the monotonicity of @xmath410 , and the continuity of @xmath411 . using again remark [ rem : limitentro ] , and the last display
, we get that the following limit exists @xmath412 where we used lemma [ lem : psiintegral ] in the last step .
this concludes the proof .
in this section we recall a general formula to compute the derivative of the conditional entropy @xmath15 with respect to noise parameters .
the formula was proved in @xcite and ( * ? ? ?
* lemma 2 ) .
we restate it in the present context and present a self - contained proof for the reader s convenience .
we consider the following setting . for @xmath20 an integer , denote by @xmath413 the set of unordered pairs in @xmath71 $ ] ( in particular @xmath414 ) .
we will use @xmath415 to denote elements of @xmath413 . for for each @xmath416
we are given a one - parameter family of discrete noisy channels indexed by @xmath417 ( with @xmath418 a non - empty interval ) , with input alphabet @xmath419 and finite output alphabet @xmath420 .
concretely , for any @xmath421 , we have a transition probability @xmath422 which is differentiable in @xmath423 .
we shall omit the subscript @xmath423 since it will be clear from the context .
we then consider @xmath424 a random vector in @xmath425 , and @xmath426 a set of observations in @xmath427 that are conditionally independent given @xmath50 .
further @xmath428 is the noisy observation of @xmath429 through the channel @xmath430 . in formulae ,
the joint probability density function of @xmath50 and @xmath103 is @xmath431 this obviously include the two - groups stochastic block model as a special case , if we take @xmath432 to be the uniform distribution over @xmath425 , and output alphabet @xmath433 . in that case
@xmath434 is just the adjacency matrix of the graph . in the following we write @xmath435 for the set of observations excluded @xmath421 , and @xmath436 for @xmath437 .
[ lemma : differentiation ] with the above notation , we have : @xmath438\big\}\ , .
\label{eq : bigderivative } \ ] ] fix @xmath439 . by linearity of differentiation ,
it is sufficient to prove the claim when only @xmath440 depends on @xmath423 .
writing @xmath441 by chain rule in two alternative ways we get @xmath442 where in the last identity we used the conditional independence of @xmath443 from @xmath50 , @xmath444 , given @xmath445 . differentiating with respect to @xmath423 , and using the fact that @xmath446 is independent of @xmath440 , we get @xmath447
consider the first term . singling out the dependence of @xmath448 on @xmath449 we get @xmath450 in the second line we used the fact that the distribution of @xmath445 is independent of @xmath423 , and the normalization condition @xmath451 .
we follow the same steps for the second term ( [ eq : derivativefirst ] ) : @xmath452\log \big[\sum_{x'_e}p_{x_e , y_e|\by_{-e}}(x'_e , y_e|\by_{-e})\big]\big\}\\ & \phantom{aa}=-\frac{\d\phantom{a}}{\d\theta}\sum_{y_e } \e\big\{\big[\sum_{x_e}p_e(y_e|x_e)p_{x_e|\by_{-e}}(x_e|\by_{-e})\big]\log \big[\sum_{x'_e}p_e(y_e|x'_e)p_{x_e|\by_{-e}}(x'_e|\by_{-e})\big]\big\}\\ & \phantom{aa}=-\sum_{x_e , y_e}\frac{\d p_e(y_e|x_e)}{\d\theta } \e\big\{p_{x_e|\by_{-e}}(x_e|\by_{-e})\log \big[\sum_{x'_e}p_e(y_e|x'_e)p_{x_e|\by_{-e}}(x'_e|\by_{-e})\big]\big\}\ , .\label{eq : derivative3 } \ ] ] taking the difference of eq .
( [ eq : derivative2 ] ) and eq .
( [ eq : derivative3 ] ) we obtain the desired formula .
we next apply the general differentiation lemma [ lemma : differentiation ] to the stochastic block model .
as mentioned above , this fits the framework in the previous section , by setting @xmath103 be the adjacency matrix of the graph @xmath49 , and taking @xmath453 to be the uniform distribution over @xmath425 . for the sake of convenience
, we will encode this as @xmath454 .
in other words @xmath455 and @xmath456 ( respectively @xmath457 ) encodes the fact that edge @xmath421 is present ( respectively , absent ) . we then have the following channel model for all @xmath439 : @xmath458 we parametrize these probability kernels by a common parameter @xmath459 by letting @xmath460 we will be eventually interested in setting @xmath461 to make contact with the setting of theorem [ thm : mainestimation ] . [ lemma : differentiationsbm ]
let @xmath16 be the mutual information of the two - groups stochastic block models with parameters @xmath462 and @xmath463 given by eq .
( [ eq : pnqnparam ] ) . then there exists a numerical constant @xmath64 such that the following happens . for any @xmath464 there exists @xmath465 such that , if @xmath466 then for all @xmath467 $ ] , @xmath468 we let @xmath469 and apply lemma [ lemma : differentiation ] .
simple calculus yields @xmath470 from eq .
( [ eq : bigderivative ] ) , letting @xmath471 , @xmath472\big\}\nonumber\\ & -\frac{1}{2}\sqrt{\frac{\op_n(1-\op_n)}{n\theta}}\sum_{e\in\pair}\sum_{x_e , y_e}x_ey_e p_{x_e}(x_e)\log p_e(y_e|x_e)\\ = & \frac{1}{2}\sqrt{\frac{\op_n(1-\op_n)}{n\theta}}\sum_{e\in\pair } \e\left\{\hx_e(\by_{-e})\log \left[\frac{\sum_{x'_e}p_e(+1|x'_e ) p_{x_e|\by_{-e}}(x'_e|\by_{-e})}{\sum_{x'_e}p_e(-1|x'_e ) p_{x_e|\by_{-e}}(x'_e|\by_{-e})}\right]\right\}\nonumber\\ & -\frac{1}{4}\sqrt{\frac{\op_n(1-\op_n)}{n\theta}}\binom{n}{2}\log\big\{\frac{p_e(+|+)p_e(-|-)}{p_e(+|-)p_e(-|+)}\big\}\ , .
\label{eq : dh } \ ] ] notice that , letting @xmath473 , @xmath474 since have @xmath475}\to 0 $ ] , and @xmath476 , we obtain the following bounds by taylor expansion @xmath477-b_0-
\frac{\delta_n}{\op_n(1-\op_n)}\,\hx_e(\by_{-e})\right|\le c\ , \frac{\theta}{\op_n(1-\op_n)n}\ , , \\
\left|\log\big[\frac{p_e(+|+)p_e(-|-)}{p_e(+|-)p_e(-|+)}\big]- \frac{2\delta_n}{\op_n(1-\op_n)}\right|\le c\ , \frac{\theta}{\op_n(1-\op_n)n}\ , , \ ] ] where @xmath478 and @xmath64 will denote a numerical constant that will change from line to line in the following .
such bounds hold for all @xmath467 $ ] provided @xmath466 . substituting these bounds in eq .
( [ eq : dh ] ) and using @xmath479 , after some manipulations , we get @xmath480 we now define ( with a slight overloading of notation ) @xmath481 , and relate @xmath482 to the overall conditional expectation @xmath483 . by bayes formula
we have @xmath484 rewriting this identity in terms of @xmath483 , @xmath485 , we obtain @xmath486 using the definition of @xmath487 , we obtain @xmath488 this in particular implies @xmath489}$ ] . from eq .
( [ eq : hxe ] ) we therefore get ( recalling @xmath490 ) @xmath491 substituting this in eq .
( [ eq : dhfinal ] ) , we get @xmath492 finally we rewrite the sum over @xmath439 explicitly as sum over @xmath493 and recall that @xmath494 to get @xmath495 since @xmath103 is equivalent to @xmath103 ( up to a change of variables ) and @xmath496 , with @xmath497 is independent of @xmath423 , this is equivalent to our claim ( recall the definition of @xmath498 , eq .
( [ eq : mmse_2vert ] ) ) . from lemma [ lemma : differentiationsbm ] and theorem [ thm : main ] , we obtain , for any @xmath499 , @xmath500 from lemma [ lem : fixedpt ] and [ lem : psiintegral ] @xmath501
y.d . and a.m. were partially supported by nsf grants ccf-1319979 and dms-1106627 and the afosr grant fa9550 - 13 - 1 - 0036 .
part of this work was done while the authors were visiting simons institute for the theory of computing , uc berkeley .
let us begin with the upper bound on @xmath502 . by using @xmath503 in eq .
( [ eq : vmsedef ] ) , we get @xmath504 where the equality on the second line follows because @xmath505 is distributed as @xmath506 .
the last inequality yields the desired upper bound @xmath507 . in order to prove the lower bound on @xmath502 assume , for the sake of simplicity , that the infimum in the definition ( [ eq : vmsedef ] ) is achieved at a certain estimator @xmath508 .
if this is not the case , the argument below can be easily adapted by letting @xmath509 be an estimator that achieves error within @xmath214 of the infimum . under this assumption , we have ,
from ( [ eq : vmsedef ] ) , @xmath510 where the last identity follows since the minimum over @xmath511 is achieved at @xmath512 .
consider next the matrix minimum mean square error .
let @xmath513}$ ] an optimal estimator with respect to @xmath502 , and define @xmath514 using eq .
( [ eq : mmsedef ] ) and the optimality of posterior mean , we obtain @xmath515}\e\big\{\big[x_ix_j-\hx_{ij}(\bg)\big]^2\big\}\\ & = \frac{1}{n^2}\e\big\{\big\|\bx\bx^{\st}-\beta(\bg)\hbx(g)\hbx^{\st}(\bg)\big\|_f^2\big\}\\ & = \e\big\{1- \frac{2\beta(\bg)}{n^2}\,\<\bx,\hbx(\bg)\>^2+\frac{\beta(\bg)^2}{n^2}\,\|\hbx(\bg)\|_2 ^ 4\big\}\\ & = 1 -\e\left(\frac{\<\bx,\hbx(\bg)\>^2}{n\|\hbx(\bg)\|_2 ^ 2}\right)^2\ , .
\label{eq : boundvmse2 } \ ] ] the desired lower bound in eq .
( [ eq : vmsebounds ] ) follows by comparing eqs .
( [ eq : boundvmse1 ] ) and ( [ eq : boundvmse2 ] ) . we shall assume , for the sake of simplicity , that the infimum in the definition of @xmath502 , see eq .
( [ eq : vmsedef ] ) is achieved for a given estimator @xmath181 .
if this is not the case , the proof below is easily adapted by considering an approximately optimal estimator .
we then define @xmath516 by letting @xmath517 notice that @xmath518 . also by the proof in previous section ,
( [ eq : boundvmse1 ] ) , we have @xmath519 and therefore ( since @xmath520 ) @xmath521 next consider the definition of overlap ( [ eq : overlapdef ] ) .
consider the randomized estimator @xmath522 defined by letting @xmath523}$ ] with @xmath524 independently across @xmath525 $ ] .
( formally , @xmath526 with @xmath527 a probability space , but we prefer to avoid unnecessary technicalities . )
we then have , by central limit theorem @xmath528 with the @xmath529 uniform in @xmath530 .
this yields the desired lower bound since , by dominated convergence , @xmath531
we prove the claim for @xmath532 ; the other claim follows from an identical argument .
since @xmath533 , we have by triangle inequality , that @xmath534 . applying bernstein inequality to the sum @xmath535 of random variables bounded by 1 : @xmath536 setting @xmath537 for large enough @xmath64 yields the required result .
let us start from point @xmath125 , .
since @xmath538 , it is sufficient to prove this claim for @xmath539 where , for the rest of the proof , we keep @xmath540 .
we start by noting that , for all @xmath541 , @xmath542 this identity can be proved using the fact that @xmath543 .
indeed this yields @xmath544 where the first and last equalities follow by symmetry . differentiating with respect to @xmath146 ( which can be justified by dominated convergence ) : @xmath545 now applying stein s lemma ( or gaussian integration by parts ) : @xmath546 using the trigonometric identity @xmath547 , the shorthand @xmath548 and identity above : @xmath549 now , let @xmath550 , whereby we have @xmath551 note now that @xmath552 satisfies @xmath54 @xmath552 is even with @xmath553 , @xmath56 @xmath552 is continuously differentiable and @xmath554 @xmath552 and @xmath555 are bounded . consider the function @xmath556 , where @xmath557 .
we have the identities : @xmath558 hence , to prove that @xmath559 is concave on @xmath281 , it suffices to show that @xmath560 , @xmath561 are non - positive for @xmath562 . by properties @xmath56 and @xmath554 above we can differentiate @xmath563 with respect to @xmath564 and interchange differentiation and expectation . computing the derivative with respect to @xmath569 yields @xmath570 where the last line follows from the fact that @xmath571 is odd and @xmath572 is even in @xmath573 . consequently @xmath574 since @xmath575 and @xmath576 for @xmath577 , the integrand is negative and we obtain the desired result .
for any random variable @xmath578 we have @xmath579 since @xmath580 ( given @xmath88 there are exactly @xmath73 possible choices for @xmath50 ) , this implies @xmath581 the claim ( [ eq : ixx ] ) follows by applying the last inequality once to @xmath582 and once to @xmath583 and taking the difference . for the second claim
, we prove that @xmath586 where @xmath587 as @xmath588 , whence the claim follows since @xmath497 .
we claim that we can construct an estimator @xmath589 and a function @xmath590 with @xmath591 , such that , defining @xmath592 then we have @xmath593 to prove this claim , it is sufficient to consider @xmath594 where @xmath595 is the principal eigenvector of @xmath103
. then @xcite implies that , for @xmath596 , almost surely , @xmath597 hence the above claim holds , for instance , with @xmath598 .
then expanding @xmath599 with the chain rule ( whereby @xmath600 ) , we get : @xmath601 since @xmath95 is a function of @xmath602 , @xmath603 .
furthermore @xmath604 since @xmath95 is binary .
hence : @xmath605 when @xmath606 , @xmath50 differs from @xmath607 in at most @xmath608 positions , whence @xmath609 .
when @xmath610 , we trivially have @xmath611 . consequently : @xmath612 the second claim then follows by dividing with @xmath20 and letting @xmath17 on the right hand side .
by definition , we have : @xmath613 define a related sequence @xmath614 as follows : @xmath615 } f'_t(s_i^t + \mu_t x_i , x(\eps)_i)\ , , \\
\bs^0 & = \bx^0 + \mu_0\bx\ , . \ ] ] here @xmath261 is defined via the state evolution recursion : @xmath616 we call a function @xmath617 is pseudo - lipschitz if , for all @xmath618 @xmath619 where @xmath620 is a constant . in the rest of the proof , we will use @xmath620 to denote a constant that may depend on @xmath236 and but not on @xmath20 , and can change from line to line .
we are now ready to prove lemma [ lem : stateevollem ] . since the iteration for @xmath621 is in the form of @xcite , we have for any pseudo - lipschitz function @xmath622 :
@xmath623 letting @xmath624 , this implies that , almost surely : @xmath625 it then suffices to show that , for any pseudo - lipschitz function @xmath626 , almost surely : @xmath627 = 0\ , .
\ ] ] we instead prove the following claims that include the above . for any @xmath236 fixed , almost surely : @xmath628 & = 0,\label{eq : induc0}\\ \lim_{n\to \infty}\frac{1}{n}\,\|\bdelta^t\|_2 ^ 2 & = 0 , \label{eq : induc1}\\ \limsup_{n\to\infty } \frac{1}{n}\,\|\bs^t + \mu_t \bx\|_2 ^ 2 & < \infty\ , , \label{eq : induc2 } \ ] ] where we let @xmath629 .
we can prove this claim by induction on @xmath236 .
the base case of @xmath630 is trivial for all three claims : @xmath631 and @xmath632 is satisfied by our initial condition @xmath633 , @xmath634 .
now , assuming the claim holds for @xmath635 we prove the claim for @xmath236 . by the pseudo - lipschitz property and triangle inequality ,
we have , for some @xmath620 : @xmath636 consequently : @xmath637}\right\rvert } & \le \frac{l}{n } \sum_{i=1}^n\left ( { \left\lvert{\delta^t_i}\right\rvert } + { \left\lvert{s^t_i + \mu_t x_i}\right\rvert } { \left\lvert{\delta^t_i}\right\rvert } + { \left\lvert{\delta^t_i}\right\rvert}^2\right)\\ & \le \frac{l}{n } \big(\|\bdelta^t\|_2 ^ 2 + \sqrt{n}\|\bdelta^t\|_2 + \|\bdelta^t\|_2\|\bs^t+\mu_t\bx\|_2\big)\ , . \ ] ] hence the induction claim eq . at @xmath236
follows from claims eq .
and eq . at @xmath236
now , with the standard inequality @xmath643 : @xmath644 using the fact that @xmath645 are lipschitz : @xmath646 by the induction hypothesis , ( specifically @xmath647 at @xmath648 , wherein it is immediate to check that @xmath649 is pseudo - lipschitz by the boundedness of @xmath650 ) : @xmath651 thus the first term in eq . vanishes .
for the second term to vanish , using the induction hypothesis for @xmath652 , it suffices that almost surely : @xmath653 this follows from standard eigenvalue bounds for wigner random matrices @xcite . for the third term in eq . to vanish , we have by @xcite that : @xmath654 hence it suffices that @xmath655 a.s . , for which we expand their definitions to get : @xmath656.\end{aligned}\ ] ] by assumption , @xmath657 is lipschitz and we can apply the induction hypothesis with @xmath658 to obtain that the limit vanishes .
indeed , by a similar argument @xmath659 is bounded asymptotically in @xmath20 , and so is @xmath660 .
along with the induction hypothesis for @xmath661 this implies that the fourth term in eq .
asymptotically vanishes .
this establishes the induction claim eq ..
e. abbe , a. s. bandeira , a. bracher , and a. singer , _ decoding binary node labels from censored edge measurements : phase transition and efficient recovery _ , ieee transactions on network science and engineering * 1 * ( 2014 ) , no . 1 .
e. abbe , a.s .
bandeira , a. bracher , and a. singer , _ linear inverse problems on erds - rnyi graphs : information - theoretic limits and efficient recovery _
, information theory ( isit ) , 2014 ieee international symposium on , june 2014 , pp .
12511255 .
mireille capitaine , catherine donati - martin , and delphine fral , _ the largest eigenvalues of finite rank deformation of large wigner matrices : convergence and nonuniversality of the fluctuations _ , the annals of probability * 37 * ( 2009 ) , no . 1 , 147 . cyril masson , andrea montanari , thomas j richardson , and rdiger urbanke , _ the generalized area theorem and some of its consequences _ , information theory , ieee transactions on * 55 * ( 2009 ) , no . 11 , 47934821 .
andrea montanari and david tse , _ analysis of belief propagation for non - linear problems : the example of cdma ( or : how to prove tanaka s formula ) _ , information theory workshop , 2006 .
itw06 punta del este .
ieee , ieee , 2006 , pp . | we develop an information - theoretic view of the stochastic block model , a popular statistical model for the large - scale structure of complex networks .
a graph @xmath0 from such a model is generated by first assigning vertex labels at random from a finite alphabet , and then connecting vertices with edge probabilities depending on the labels of the endpoints . in the case of the symmetric two - group model
, we establish an explicit ` single - letter ' characterization of the per - vertex mutual information between the vertex labels and the graph . the explicit expression of the mutual information is intimately related to estimation - theoretic quantities , and in particular reveals a phase transition at the critical point for community detection . below the critical point
the per - vertex mutual information is asymptotically the same as if edges were independent .
correspondingly , no algorithm can estimate the partition better than random guessing .
conversely , above the threshold , the per - vertex mutual information is strictly smaller than the independent - edges upper bound . in this regime
there exists a procedure that estimates the vertex labels better than random guessing . |
pair diffusion features prominently in theories of reaction - diffusion dynamics@xcite describing processes where reactant encounters are required , such as ligand binding and aggregation or fluorescence quenching .
the hydrodynamic interactions quantified by the distance - dependent diffusion coefficient are also central to the theory and simulation of polymer dynamics , including protein folding simulations in implicit solvent , the hydrodynamic coupling in dense colloidal suspensions , and the function of nanomachines and bacterial flagella.@xcite considering the broad importance of pair diffusion in theories of molecular kinetics , it may seem surprising that little is known about the pair diffusion coefficient and its dependence on the particle distance .
formidable challenges in both theory and simulations@xcite have resulted in often contradictory results for this fundamental quantity .
theoretically , the pair diffusion coefficient @xmath0 ( with @xmath1 the distance between two particles ) has been attacked from two opposite directions , building up from kinetic theory@xcite or projecting down from macroscopic hydrodynamics.@xcite for @xmath0 , kinetic theory had limited success at high fluid packing densities , largely because of the complexity of the molecular motions in dense fluids resulting from their many - body character . at the other extreme ,
details of the molecular structure of the solvent are ignored in estimates of the pair friction derived from macroscopic hydrodynamics , for instance by using the oseen or rotne - prager tensors.@xcite nevertheless , this approach has proved useful in studies of the dynamics of large and sufficiently distant pairs of colloidal particles in a solvent,@xcite where macroscopic hydrodynamics is expected to apply ; but it is not immediately applicable when solute and solvent particles are of comparable size , for instance in ( aqueous ) solutions of ( bio)polymers . here , we determine the pair diffusion coefficient directly from the simulated many - body dynamics in a dense fluid .
we focus on particles of the same size as the solvent molecules .
this small - solute regime is of particular relevance because , on the one hand , it allows us to quantify hydrodynamic interactions relevant for molecular motions , including the dynamics of ( bio)polymers in solution , and , on the other hand , it is far outside the regime where macroscopic hydrodynamics should be expected to apply .
the paper is outlined as follows . in section
i , we describe the methodological details , including the theory to calculate the pair diffusion tensor , the algorithm used to determine the required green s functions from simulation data , the simulation parameters , and the validation procedure .
we validate our method by computing the pair diffusion coefficient for two spherical particles subject to brownian dynamics . in the results section
ii , we first present a comparison of green s functions obtained from simulations against those predicted from our diffusion model , finding excellent agreement over 8 orders of magnitude .
then we examine the pair diffusion coefficients as a function of distance between two particles for several fluid packing fractions , and compare the simulation results to the predictions of macroscopic hydrodynamic theories .
finally , we show that the position - dependent pair diffusion coefficient is correlated to the local available volume . in the appendix
, we discuss the calculation of the angular pair diffusion coefficient .
in the following we present the theory to calculate the position - dependent pair diffusion tensor from simulation trajectory data .
the diffusion tensor @xmath4 of the vector @xmath5 between two given particles in an isotropic and homogeneous fluid has spherical symmetry : @xmath6 where @xmath7 is the length of the pair vector ; @xmath8 and @xmath9 are the scalar diffusion coefficients in the radial and tangential directions , respectively ; and @xmath10 , @xmath11 , and @xmath12 are the orthonormal unit vectors of the spherical polar coordinate system , with @xmath10 pointing in the radial direction , and @xmath11 and @xmath12 being tangential to longitudes and latitudes , respectively .
the smoluchowski ( or fokker - planck ) equation describing the diffusion of the pair vector then takes on the following form : @xmath13\end{aligned}\ ] ] where @xmath14 is the green s function for a pair vector starting at a distance @xmath15 and azimuthal angle @xmath16 , without loss of generality because of the isotropic space ( making the @xmath17 distribution uniform ) ; @xmath18 is the distance - dependent free energy surface ; @xmath19 is the inverse temperature ; @xmath20 is the partial derivative with respect to time ; and `` div '' and `` grad '' are the divergence and gradient operators in spherical polar coordinates , respectively .
we will in the following use @xmath21 instead of @xmath22 .
let @xmath23 be the green s function in terms of this new variable .
the diffusion equation eq .
( [ eq : smolu ] ) then becomes : @xmath24}\nonumber\\ & & + \frac{d_\parallel(r)}{r^2}\partial_x\left[(1-x^2)\partial_xp\right]\end{aligned}\ ] ] where @xmath25 . by integrating over @xmath21
we obtain a diffusion equation for the green s function in the radial direction alone , with the second term on the right hand side vanishing : @xmath26~,\end{aligned}\ ] ] where @xmath27 is the probability for the pair distance to be in @xmath28 at time @xmath29 , starting from @xmath30 at time 0 . as a consequence ,
we can treat radial diffusion separately using standard one - dimensional diffusion , irrespective of the angular motion . in an appendix
, we outline an extension of the theory to the orientational diffusion of the pair distance vector . here
we focus on the calculation of the position - dependence of the pair - distance diffusion coefficient @xmath31 , where we have dropped the subscript for notational simplicity . in our calculations of @xmath0
, we face the dual challenges that it depends on the particle distance @xmath1 , and that the pair dynamics becomes diffusive only at times at which the influence is felt of the underlying free energy surface ( or potential of mean force ) , @xmath32 , where @xmath33 is the pair correlation function of the two particles in the fluid . to disentangle the diffusive spread of the pair distance distribution from the drift of the mean position as a result of the underlying free energy surface , we use the propagator ( or green s function ) @xmath34 . in constructing a diffusion model , we assume that @xmath35 satisfies the smoluchowski diffusion equation eq .
( [ eq : dr ] ) , where the term within the brackets is the negative of the radial probability flux . a spatial discretization of the smoluchowski equation@xcite results in a master equation that describes the pair dynamics between neighboring intervals along @xmath1 .
the particle - pair trajectories in the simulations are discretized by assigning pair distances into bins @xmath36 along @xmath1 , and then counting the numbers @xmath37 that a pair distance is in bin @xmath36 at time @xmath38 , and in bin @xmath39 at time @xmath40 , irrespective of its location at intervening times , with @xmath41 the lag time .
@xmath37 is symmetrized , @xmath42 , consistent with microscopic time reversibility .
we then find the pair diffusion coefficient @xmath0 that maximizes the path action of the observed discretized path .
for the discretized diffusion model with given @xmath43 and @xmath44 , the path action ( or likelihood ) @xmath45 can be be written as a product of green s functions that are expressed in terms of a matrix exponential.@xcite to optimize the action and find the diffusion model most consistent with the observed @xmath37 , we infer @xmath43 and @xmath44 using a bayesian approach,@xcite with uniform priors in @xmath46 and @xmath44 ensuring scale invariance in time and space . in free diffusion ,
one typically fits @xmath47 ( or , equivalently , @xmath48 ) to the mean - square displacement , with the constant @xmath49 ( or the time shift @xmath50 ) accounting for initial fast molecular motions . here , we employ a similar procedure by optimizing also the time origin @xmath38 for transition counts @xmath51 collected at several different lag times @xmath52 , where @xmath29 defines the `` observation time . ''
to validate the procedure , we first run brownian dynamics simulations for two spherical particles of unit diameter freely diffusing with diffusion coefficient @xmath53 in a cubic box of length @xmath54 under periodic boundary conditions and with reflecting boundaries at particle contact . by construction , in this case @xmath55 , which is indeed recovered by the procedure for distances @xmath56 ( fig .
[ brownian ] ) , nearly independent of grid size @xmath57 .
however , for @xmath58 and long lag times , the periodic boundary conditions cause artifacts because in the corners of the cubic simulation box the pair dynamics projected onto the minimum image distance depends not only on the length of the pair vector but also on its direction .
to calculate @xmath0 for a particle pair in a dense fluid , we perform discontinuous molecular dynamics ( dmd ) simulations of hard sphere ( hs ) fluids . in dmd , particles follow linear trajectories between collisions . in a collision , the velocities of colliding particles are changed to conserve energy and momentum.@xcite to simplify the notation , dimensionless quantities will be used , obtained by appropriate combinations of a characteristic length ( hs particle diameter @xmath59 ) and time scale ( @xmath60 , where @xmath61 is the particle mass ) .
the packing fraction @xmath62 is defined in terms of the particle density @xmath63 . to construct the green s functions , we performed dmd simulations with @xmath64 identical hs particles .
periodic boundary conditions were applied in all directions .
the average self - diffusivity @xmath65 was obtained by fitting the long - time ( @xmath66 ) behavior of the average mean - squared displacements @xmath67 of the particles to the einstein relation @xmath68 . to minimize the system - size dependence,@xcite trajectories from simulations with @xmath69 particles
were used to determine @xmath65 , with remaining finite - size corrections of @xmath701 % .@xcite
to test the applicability of the diffusion model , we compare its prediction for the dynamics of the pair distance to actual simulation data collected over a range of time scales .
figure [ greens ] shows that diffusion quantitatively captures the pair dynamics in the fluid .
the green s functions @xmath71 from the diffusion model and the results of the dmd simulation data are found to agree over 8 orders of magnitude . at the shortest observation time @xmath72
, we find that the green s functions are essentially gaussian with position - dependent widths . at longer times , @xmath73 and 20
, the underlying free energy surface shows its influence , distorting the propagators away from the gaussian form expected for free diffusion on a flat surface . in figure [ param ]
, we explore the effects of the spatial grid size @xmath57 and the observation time @xmath29 on the calculated pair diffusion coefficient .
we find that for @xmath74 , grid size effects are negligible .
figure [ param ] ( bottom ) shows that the effect of changing the observation time @xmath29 is negligible only for shorter distances @xmath75 .
in contrast , for longer distances @xmath0 is almost flat at a short observation time @xmath76 and does not show the asymptotic @xmath3 dependence expected from macroscopic hydrodynamic theory .
however , the expected @xmath3 dependence is recovered for longer times @xmath29 .
this result implies that the hydrodynamic coupling at large distances is not instantaneous , such that a more accurate diffusion model would require the inclusion of memory effects in a frequency and position - dependent diffusion coefficient.@xcite for @xmath77 the predictions are essentially independent of @xmath29 . in all following calculations ,
we thus use @xmath78 and @xmath79 .
+ having validated the procedure and diffusion model , we now examine the distance - dependent pair diffusion coefficients @xmath0 for different packing fractions @xmath80 .
figure [ dr ] ( top panel ) shows @xmath0 for the hs fluid over a packing fraction range @xmath81 ( symbols from top to bottom ) .
also shown are the predictions for @xmath0 from hydrodynamic theory for two spherical particles with slip boundary conditions,@xcite as well as the widely - used oseen tensor correction@xcite ( for @xmath82 ; dashed line ) , which for the pair diffusion coefficient is @xmath83 where @xmath84 is the solvent shear viscosity , taken from ref . .
we find that both the exact hydrodynamic theory and the oseen approximation ( and similarly the rotne - prager tensor;@xcite not shown ) are remarkably accurate and quantitatively reproduce the large-@xmath1 behavior .
however , hydrodynamic predictions only qualitatively reproduce the observed decrease in @xmath0 near contact ( @xmath85 ) and lack any structure due to molecular correlations in the first- and second - shell around a particle . to characterize the effects of the molecular packing structure on the pair dynamics , we plot in fig .
[ dr ] ( bottom panel ) the normalized pair diffusion coefficient @xmath86 for different packing fractions @xmath80 . as expected from macroscopic hydrodynamics , at large distances @xmath1 the @xmath86 data collapse onto a single curve that is well represented by the hydrodynamic theory .
two important observations are : ( i ) @xmath86 is always less than 1 , with pair diffusion slowed down by `` hydrodynamic interactions . ''
( ii ) @xmath0 rises sharply just outside distances of 1 and 2 particle diameters , and drops sharply just outside @xmath87 , and 2.5 .
this strong position dependence , together with the short - time propagator @xmath88 for brownian dynamics being gaussian with mean @xmath89 $ ] , implies that at short times particle pairs just outside the first and second shell boundary `` drift '' outward , whereas those inside the boundary drift inward , beyond what is expected from the free energy gradient @xmath90 alone .
the additional drift terms arise from the large gradients in @xmath0
. we can understand this dynamic behavior ( which does not violate microscopic time reversibility and detailed balance ! ) from the many - body packing effects . at @xmath91 , for instance
, the interstitial space between the two particles is likely filled by a third one , which tends to drive the pair apart .
in contrast , at @xmath92 , the interstitial space between the two particles is empty , and the two particles tend to move closer together . to gain further insight into the observed structure in @xmath0 and its relation to the static structure of the fluid , we plot in figure [ gr - dr]a both @xmath0 and the pair correlation function @xmath33 .
we find that there is some correlation between the structure in @xmath0 and @xmath33 except near the contact distance at @xmath85 where these quantities are actually anti - correlated .
somewhat counter - intuitively , this mostly positive correlation means that the pair diffusion is actually higher in the more densely packed regions .
similar behavior was observed for a hs fluid confined between hard walls where the local density was found to be strongly correlated with the local diffusion coefficient except near the walls.@xcite this behavior was found to be related to the physics of layer formation , with the available volume , as probed by the local test - particle insertion probability @xmath93 , being largest in the locally dense regions of space.@xcite a similar argument should hold in our case of a bulk hs fluid in which purely entropic excluded volume forces give rise to a structured @xmath33 profile to maximize the system entropy .
the local insertion probability is given by @xmath94 , where the activity @xmath95 is spatially invariant for an equilibrium fluid , with @xmath96 the chemical potential , and @xmath97 the thermal wavelength .
+ to test if @xmath0 is indeed related to @xmath98 , we calculate @xmath99 for the different packing fractions by utilizing grand canonical transition - matrix monte carlo simulations.@xcite figure [ gr - dr]b shows @xmath0 versus @xmath98 for different @xmath80 .
we find that the @xmath0 data approximately collapse onto a curve similar to the average bulk relationship ( @xmath100 versus @xmath93 ) that ignores any @xmath1 dependence .
therefore , at least as a rough approximation , the local available volume can describe the pair diffusion in this case .
the results of this paper shed light on the microscopic origins of the distance dependence of hydrodynamic interactions , in particular the role of particle packing and many - body motions , and help establish a range of validity for the assumption of macroscopic hydrodynamics in the modeling of processes ranging from polymer dynamics to nanomachines , colloidal dynamics , and bacterial swimming . in practical applications , such as the calculation of diffusional encounter rates , the significant deviations between the calculated pair diffusion coefficients @xmath0 and the ideal ( and widely used ! ) assumption of @xmath101 can result in substantial errors , with @xmath102 consistently . at
the least one should use a hydrodynamic theory , with both the exact theory and the oseen tensor giving remarkably accurate results for hydrodynamic interactions at larger distances , and rough approximations in the regime dominated by molecular packing near contact .
to treat the angular diffusion of pair distance vectors ( or other vectors in an isotropic space ) , we notice that the second term on the right hand side of eq .
( [ eq : smolux ] ) corresponds to the angular momentum operator in quantum mechanics .
we thus make the ansatz @xmath103 , where the @xmath104 are the legendre polynomials of order @xmath105 , and the coefficients @xmath106 do not depend on @xmath29 and @xmath1 . with this ansatz
, we obtain uncoupled one - dimensional evolution equations for each of the @xmath107 ( with @xmath108 ) : @xmath109 - \frac{d_\parallel(r)}{r^2}l(l+1)q_l~.\ ] ] for @xmath110 , this expression is identical to eq .
( [ eq : dr ] ) ; for @xmath111 , these are sink ( or birth - death ) equations for the @xmath107 , with sink terms whose strength increases quadratically with @xmath105 , and with @xmath112 .
that is , at long times only the distribution uniform in @xmath113 survives ( with @xmath114 )
. expressed in terms of dirac @xmath115-functions , the initial condition for the green s function is @xmath116 ( where we chose the coordinate system such that the polar axis points in the direction of the pair distance vector at time zero ) , with normalization @xmath117 . by using the orthogonality relations of the legendre polynomials , @xmath118 with @xmath119 the kronecker-@xmath115 , we obtain @xmath120 where the @xmath107 satisfy eq .
( [ eq : dql ] ) with initial conditions @xmath121 . for the sake of completeness
, we also sketch an algorithm to obtain the distance - dependent radial and angular diffusion coefficients @xmath8 and @xmath9 from simulation data ( or , equivalently , from experimental data , such as those obtained in colloidal - particle tracking experiments ) .
1 . use counts of transitions @xmath37 from bins @xmath36 to @xmath39 in the radial direction only ( irrespective of the angular motion ) as input in the algorithm@xcite described above to calculate the one - dimensional position - dependent diffusion coefficients @xmath8 , and the potential of mean force @xmath18 .
2 . determine counts @xmath122 for transitions from bin @xmath36 in the radial direction to bin @xmath123 in a two dimensional histogram .
radial bins are indexed by @xmath39 , and angular bins by @xmath124 according to the cosine of the azimuthal angle , @xmath125 ( with @xmath126 and @xmath127 by definition of the coordinate system ) .
3 . with @xmath8 and @xmath18 already determined in the first step , the green s function eq .
( [ eq : p ] ) can be calculated for a given estimate of @xmath9 from a spatially discretized version@xcite of the sink equations , eq .
( [ eq : dql ] ) . with this green s function ,
one can again use a bayesian inference procedure ( or maximum - likelihood method ) to estimate the @xmath9 ( on lattice points halfway between the bin centers ) that is most consistent with the observed transition counts @xmath122 .
note that the infinite sum over @xmath105 in eq .
( [ eq : p ] ) has to be truncated in practical calculations .
note further that the same algorithm can also be used to determine the diffusion coefficients of a single particle in confinement with spherical symmetry .
we thank dr .
attila szabo for many helpful discussions .
this research was supported by the intramural research program of the nih , niddk , and utilized the high - performance computational capabilities of the biowulf pc / linux cluster at the national institutes of health , bethesda , md ( http://biowulf.nih.gov ) .
19ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we calculate the pair diffusion coefficient @xmath0 as a function of the distance @xmath1 between two hard - sphere particles in a dense monodisperse suspension .
the distance - dependent pair diffusion coefficient describes the hydrodynamic interactions between particles in a fluid that are central to theories of polymer and colloid dynamics .
we determine @xmath0 from the propagators ( green s functions ) of particle pairs obtained from discontinuous molecular dynamics simulations . at distances exceeding @xmath23 molecular diameters ,
the calculated pair diffusion coefficients are in excellent agreement with predictions from exact macroscopic hydrodynamic theory for large brownian particles suspended in a solvent bath , as well as the oseen approximation .
however , the asymptotic @xmath3 distance dependence of @xmath0 associated with hydrodynamic effects emerges only after the pair distance dynamics has been followed for relatively long times , indicating non - negligible memory effects in the pair diffusion at short times .
deviations of the calculated @xmath0 from the hydrodynamic models at short distances @xmath1 reflect the underlying many - body fluid structure , and are found to be correlated to differences in the local available volume .
the procedure used here to determine the pair diffusion coefficients can also be used for single - particle diffusion in confinement with spherical symmetry . |
measurements of the expansion rate history @xmath11 of the universe , when interpreted within the standard model of cosmology , convincingly indicate that the universe has recently entered a phase of accelerated expansion @xcite .
most of this unexpected evidence is provided via geometric probes of cosmology , that is by constraining the redshift scaling of the luminosity distances @xmath15 of cosmic standard candles ( such as supernovae ia ) , or of the angular diameter distance @xmath16 of cosmic standard rulers ( such as the sound horizon scale at the last scattering epoch ) . despite much observational evidence ,
little is known about the physical mechanism that drives cosmic acceleration . as a matter of fact ,
virtually all the attempts to make sense of this perplexing phenomenon without invoking a new constant of nature ( the so called cosmological constant ) call for exotic physics beyond current theories .
for example , it is speculated that cosmic acceleration might be induced by a non clustering , non evolving , non interacting and extremely light vacuum energy @xmath3 @xcite , or by a cosmic field with negative pressure , and thus repulsive gravitational effect , that changes with time and varies across space ( the so called dark energy fluid ) @xcite , if not by a break - down of einstein s theory of gravity on cosmological scales ( the so called dark gravity scenario ) @xcite .
this last , extreme eventuality is made somewhat less far - fetched by the fact that a large variety of nonstandard gravitational models , inspired by fundamental physics arguments , can be finely tuned to reproduce the expansion rate history of the successful standard model of cosmology , the @xmath3cdm paradigm .
although different models make undistinguishable predictions about the amplitude and scaling of background observables such as @xmath17 and @xmath18 the analysis of the clustering properties of matter on large linear cosmic scales is in principle sufficient to distinguish and falsify alternative gravitational scenarios .
indeed , a generic prediction of modified gravity theories is that the newton constant @xmath19 becomes a time ( and possibly scale ) dependent function @xmath20 .
therefore , dynamical observables of cosmology which are sensitive to the amplitude of @xmath19 , such as , for example , the clustering properties of cosmic structures , provide a probe for resolving geometrical degeneracies among models and for properly identifying the specific signature of nonstandard gravitational signals .
considerable phenomenological effort is thus devoted to engineering and applying methods for extracting information from dynamical observables of the inhomogeneous sector of the universe @xcite .
indeed , thanks to large and deep future galaxy redshift surveys , such as for example euclid @xcite , the clustering properties of matter will be soon characterized with a ` background level ' precision , thus providing us with stringent constraints on the viability of alternative gravitational scenarios . extending the perimeter of precision cosmology beyond zeroth order observables into the domain of first order perturbative quantities critically depends on observational improvements but also on the refinement of theoretical tools . among the quantities that are instrumental in constraining modified gravity models , the linear growth index @xmath12 , @xmath21 where @xmath22 is the scale factor of the universe
, @xmath23 is the reduced density of matter and @xmath24 the dimensionless density contrast of matter , has attracted much attention . despite being in principle a function ,
this quantity is often , and effectively , parameterized as being constant @xcite . among the various appealing properties of such an approximation , two in particular
deserve to be mentioned .
first , the salient features of the growth rate history of linear structures can be compressed into a single scalar quantity which can be easily constrained using standard parameter estimation techniques . as it is the case with parameters such as @xmath25 , @xmath26 , etc . ,
which incorporate all the information contained in the expansion rate function @xmath11 , so it is extremely economic to label and identify different growth histories @xmath27 with the single book - keeping index @xmath12 .
moreover , since the growth index parameter takes distinctive values for distinct gravity theories , any deviation of its estimated amplitude from the reference value @xmath28 ( which represents the exact asymptotic early value of the function @xmath29 in a @xmath3cdm cosmology @xcite ) is generically interpreted as a potential signature of new gravitational physics .
however useful in quantifying deviations from standard gravity predictions , this index must also be precise to be of any practical use . as a rule of thumb , the systematic error introduced by approximating @xmath29 with @xmath30 , which depends on @xmath31 ,
must be much smaller than the precision with which future experiments are expected to constrain the growth index over a wide redshift range(@xmath32 @xcite ) .
notwithstanding , already within a standard @xmath3cdm framework with @xmath33 , the imprecision of the asymptotic approximation is of order @xmath34 at @xmath35 .
more subtly , the expansion kinematic is expected to leave time dependent imprints in the growth index .
the need to model the redshift evolution of the growth index , especially in view of the large redshift baseline that will be surveyed by future data , led to the development of more elaborated parameterizations @xcite . whether their justification is purely phenomenological or theoretical , these formulas aim at locking the expected time variation of @xmath29 into a small set of scalar quantities , the so called growth index parameters @xmath1 . for example
, some authors ( e.g. @xcite ) suggest to use the taylor expansion @xmath36_{z=0}\ , z$ ] for data fitting purposes . indeed , this approach has the merit of great accuracy at present epoch , but it becomes too inaccurate at the intermediate redshifts ( @xmath37 ) already probed by current data . on top of precision issues , there are also interpretational concerns . ideally , we would like the growth index parameter space to be also in one - to - one correspondence with predictions of specific gravitational theories .
in other terms we would like to use likelihood contours in this plane to select / reject specific gravitational scenarios .
this is indeed a tricky issue .
for example , it is rather involved to link the amplitude of the growth index parameters to predictions of theoretical models if the growth index fitting formula has a phenomenological nature .
more importantly , it is not evident how to extract growth information ( @xmath38 ) from a function , @xmath12 which , as equation ( [ defuno ] ) shows , is degenerate with background information ( specifically @xmath39 ) .
in other terms , the growth index parameters are model dependent quantity that can be estimated only after a specific model for the evolution of the background quantity @xmath39 is supplied .
therefore it is not straightforward to use the likelihood function in the @xmath12-plane to reject dark energy scenarios for which the background quantities do not scale as in the fiducial . because of this , up to now , growth index measurements in a given fiducial
were used to rule out only the null - hypothesis that the fiducial correctly describes large scale structure formation processes .
growth index estimates were not used to gauge the viability of a larger class of alternative gravitational scenarios .
a reverse argument also holds and highlights the importance of working out a growth index parameterization which is able to capture the finest details of the exact numerical solution , establishing at the same time , the functional dependence on background observables .
indeed , once a given gravitational paradigm is assumed as a prior , the degeneracy of growth index measurements with background information , can be exploited to constraining the background parameter of the resulting cosmological model , directly using growth data .
therefore , by expressing the growth index as a function of specific dark energy or dark gravity parameters one can test for the overall coherence of cosmological information extracted from the joint analyses of smooth and inhomogeneous observables . in this paper
we address some of these issues by means of a new parameterization of the growth index .
the main virtues of the approach is that the proposed formula is _ a ) _ flexible , i.e. it describes predictions of a wide class of cosmic acceleration models , _ b ) _ accurate , i.e. it performs better than alternative models in reproducing exact numerical results , _ c ) _ it is observer friendly , i.e. accuracy is achieved with a minimum number of parameters and _ d ) _ it is ` theorist friendly ' , i.e. the amplitude of the fitting parameters can be directly and mechanically related , in an analytic way , to predictions of theoretical models .
the paper is organized as follows .
we define the parameterization for the growth index in section [ sec : parametrizing ] , and we discuss its accuracy in describing various dark energy models such as smooth and clustering quintessence in [ sec : standard ] . in [ sec : modified ] we apply the formalism to modified gravity scenarios . in particular , we discuss the dgp @xcite and the parameterized post friedmanian @xcite scenarios . in [
sec : constraining ] we will impose the studied models to current ( simulated future ) data .
conclusions are drawn in [ sec : conclusions ] . throughout all the paper ,
if not specified differently , the flat friedmann - lematre - robertson - walker cosmology with @xmath4 parameters @xmath40 @xcite is referred to as the _ reference _ @xmath3cdm model .
in this section we introduce our notation and we give the theoretical background for analyzing the clustering of matter on large linear scales . for a large class of dark energy models and gravitational laws , at least on scales where the quasi - static approximation applies , the dynamics of linear matter perturbations @xmath41 can be effectively described by the following second order differential equation @xmath42 where overdot ( @xmath43 ) denotes derivation with respect to cosmic time @xmath44 and the dimensionless response @xmath45 and damping @xmath46 coefficients are general functions of cosmic time and possibly the spatial fourier frequency . as for the simplest case ,
general relativity even augmented by a minimally coupled scalar field , marks the point @xmath47 while for modified gravity models we expect , in general , @xmath48 .
the specific form of @xmath49 predicted in higher dimension brane models @xcite or in scalar tensor theories of gravity @xcite are given in [ sec : dgp ] and [ sec : skofe ] respectively .
see , instead , @xcite for a more elaborate model of modified gravity which does not reduce to the standard form of eq .
[ eq : matter_density_fluctuations ] .
it is convenient , from the observational perspective , to express eq .
[ eq : matter_density_fluctuations ] in terms of the linear growth rate @xmath50 , a cosmic observable defined as the logarithmic derivative of the matter density fluctuations @xmath51 .
we obtain @xmath52 where prime ( @xmath53 ) denotes derivation with respect to @xmath54 .
it is standard practice to look for solutions of eq .
[ eq : f_h_general ] with the form @xmath55 where @xmath12 is called the growth index already introduced in [ sec : introduction ] .
although , in some special cases , an exact analytic solution can be explicitly given ( for example , in terms of a hypergeometric function if @xmath56 and the de _ eos _ is constant @xcite ) , the approximate parametric solution @xmath57 has been widely advocated @xcite .
indeed , the growth rate of matter density perturbations is highly sensitive to the kinematics of the smooth expansion of the universe , as well as to the gravitational instability properties of the background fluids .
the approximate anstaz @xmath57 interestingly and neatly splits the background ( basis ) and the inhomogeneous contribution ( exponent ) to the linear growth rate of structures .
gravitational models beyond standard general relativity , which predict identical background expansion histories , i.e. identical observables @xmath58 and @xmath59 , may thus be singularized and differentiated by constraining the amplitude of @xmath12 . however solutions with a constant growth index of ( sub - horizon ) matter density perturbations can not be realized in quintessence models of dark energy over the whole period from the beginning of matter domination up into distant future .
moreover , it is likely that different gravitational models tuned to reproduce the same expansion history @xmath60 display unidentical evolution for @xmath61 .
this implies that the amplitudes of @xmath12 inferred in two different cosmological models , characterized by distinctive evolution laws of the matter density parameter , can not be directly compared . in order to increase the accuracy of predictions , we look for a more elaborate parametric form of the growth index : we express @xmath12 as a function of @xmath62 .
plugging ( [ eq : f_approximation ] ) into ( [ eq : f_h_general ] ) we obtain @xmath63 where we set @xmath64 .
we assume that all non - constant coefficients ( @xmath65 , @xmath66 , @xmath49 , @xmath46 ) are well - defined functions of @xmath67 , and are completely specified once a theory of gravity is considered .
since the numerical solution of equation ( [ eq : f_h_x ] ) for a @xmath3cdm model indicates that @xmath12 is an almost linear function of @xmath67 , and guessing that viable alternative theories of gravity should predict minimal , but detectable , deviations from standard model results , we suggest to describe the time evolution of the growth index as @xmath68 where @xmath69 are constant parameters . expressing @xmath12 via a series expansion
has already been proposed earlier by @xcite and improved later by @xcite . in those works @xmath70 ,
whereas here we set @xmath64 .
demonstrating the gain in accuracy achieved by this change of variable is one of the goals of this paper .
a different one , but equally important , is to show that by this choice we can work out a closed analytic formula that predicts the amplitude of the coefficients @xmath71 ( up to an arbitrary order @xmath72 ) once any given dark energy and gravitational model is specified .
specifically , we define the structural parameters of the formalism @xmath73_{{\omega}=0 } \ ; , \quad \mathcal{h}_{n } : = \big[\tfrac{d^n}{d{\omega}^n}\big(\tfrac{h'}{h}\big)\big]_{{\omega}=0 } \ ; , \quad \mathcal{m}_{n } : = \big[\tfrac{d^n}{d{\omega}^n}\big(\mu \big)\big]_{{\omega}=0 } \ ; , \quad \mathcal{n}_{n } : = \big[\tfrac{d^n}{d{\omega}^n}\big(\nu \big)\big]_{{\omega}=0 } \ ] ] where @xmath74 is a natural number and @xmath75 .
we obtain ( see [ sec : annexe ] ) @xmath76 for @xmath77 , we have @xmath78 where @xmath79 and the bell polynomials are defined by @xmath80 where the sum is taken over all @xmath81-tupels @xmath82 of non - negative integers satisfying @xmath83 in the following we will test the precision of the approximation ( [ eq : gamma_taylor ] ) for different accelerating models and for various cosmological parameters .
in particular we will show that two coefficients @xmath30 and @xmath84 are sufficient for a large class of models .
in this section we frame our analysis in a friedmann lematre roberston & walker ( flrw ) model of the universe , a metric model characterized by two degrees of freedom : the spatial curvature of the universe @xmath81 and the scale factor @xmath85 .
we therefore assume that gravity is ruled by the standard einstein field equations , and we further assume that the hyper - surfaces of constant time are flat ( @xmath86 ) as observations consistently suggest @xcite , and that , at least during the late stage of its evolution , i.e. the period we are concerned with , the universe is only filled with cosmic matter and dark energy .
these are perfect fluid components whose @xmath87 , i.e. the ratio @xmath88 between pressure and density , is zero and lower than @xmath89 respectively .
we refer to the particular case in which the de @xmath87 @xmath90 as to the @xmath3cdm model .
the evolution of the physical density of matter and de is given by @xmath91 where @xmath92 while the expansion rate of the cosmic metric , or simply the hubble parameter @xmath58 , is given by @xmath93 \label{eq : friedmann}\ ] ] where @xmath26 is the present day reduced density of matter . in this notation , the evolution equations for the reduced density of matter and de are @xmath94 and @xmath95 respectively .
under these hypotheses , the relevant quantities in eq .
( [ eq : f_h_x ] ) become @xmath96 where @xmath97 is the ( possibly varying ) @xmath87 parameter of the dark energy fluid .
we will illustrate the performances of the proposed parameterization ( cf .
( [ eq : gamma_taylor ] ) ) first by assuming that dark energy is a smooth component which does not cluster on any scale , i.e. its energy density is a function of time only
. we will then increase the degrees of freedom associated to this hypothetical component , by assuming that dark energy may become gravitationally unstable , that is its effective density varies also as a function of space .
the propagation of instabilities in the energy density and pressure of a fluid can be heuristically understood in terms of two kinematical quantities : the speed of sound @xmath98 and the sound horizon @xmath99 . on scales larger than the sound horizon instabilities can collapse . on scales below @xmath100
the fluid is smooth . as an example of this last case
, one can consider a canonical , minimally coupled , scalar field such as quintessence .
since @xmath101 , the sound horizon coincides with the cosmic horizon and the quintessence fluid can be considered homogeneous on all ( observationally ) relevant scales . in our notation , such a smooth dark energy component is effectively described by setting @xmath102 into eq .
( [ eq : f_h_x ] ) .
setting @xmath103_{{\omega}=0}\ ; \label{eq : gamma_coeff_w}\end{aligned}\ ] ] we find @xmath104 for @xmath105 etc .
smooth models of dark energy make physical sense only for @xmath106 .
null energy condition _
, indeed , prevents both ghost and gradient instabilities ( e.g. @xcite ) .
notwithstanding , sound theories of de can be constructed also by relaxing the null energy condition , for example , by imposing the lagrangian density of the gravitational field to contain space derivatives higher than two , i.e. terms that become important at some high momentum scale ( see 2.2 of ref .
@xcite ) . in
what follows , we will thus consider also super acceleration scenarios @xmath107 , their effective , phenomenological character being understood .
we now discuss two distinct scenarios .
first , we assume that the dark energy _ eos
_ varies weakly with cosmic time , i.e. that it can be effectively described in terms of a constant parameter @xmath108 .
then we complete our analysis by computing the explicit amplitude of the growth index parameters in a scenario in which the de _ eos _ is free to vary .
cdm model .
the precision is estimated as the relative difference with respect to the numerical reconstruction of the growth index @xmath0 in the reference model . ] in this case , in eq .
[ eq : gamma_coeff_w ] we have @xmath109 and @xmath110 for @xmath77 .
we therefore obtain from eq .
( [ eq : gamma_coeff_x_h ] ) @xmath111 where @xmath105 etc . replacing ( [ eq : coeff_gr_1 ] ) in ( [ eq : gamma_0 ] ) and ( [ eq : gamma_n ] )
we find : @xmath112 in particular , for the @xmath3cdm model ( i. e. for @xmath90 ) , we find @xmath113 for @xmath33 , the relative error @xmath114 at order 1 , @xmath115 at order 2 and @xmath116 at order 3 .
more generally , figure [ fig : wcdm ] ( upper right panel ) shows that the accuracy of the approximation increases by roughly a factor of 10 as soon as a new coefficient is switched on . this same figure ( lower right panel )
shows that the relative accuracy depends only slightly on the equation of state parameter @xmath108 , although the error is in general slightly larger for larger @xmath108 .
the relative error of the approximation is also quite stable as we vary @xmath26 within a sensitive range of values @xmath117 $ ] ( see the lower left panel in figure [ fig : wcdm ] ) .
we remark that the coefficients @xmath30 and @xmath84 are the same ] when expanding the growth index in powers of @xmath118 whereas @xmath119 and higher orders differ . however , as a consequence of developing in @xmath62 , already at first order our approximation is , depending on cosmology , from 5 to 10 times better at @xmath35 .
this is seen in figure [ fig : comparison ] , where we also compare the accuracy of eq . with various other parameterizations of the growth index .
not only eq .
is more precise than formulas based on perturbative expansions around @xmath120 , in the relevant redshift range for dark energy studies , that is @xmath121 , it is also more accurate than the phenomenological @xcite or perturbative @xcite expressions normalized at @xmath35 .
parameters ( @xmath122 ) , in approximating the growth index of smooth dark energy models with the first order model @xmath123 of @xcite .
_ middle panel _ : accuracy of the phenomenological fitting formula of @xcite .
_ right panel _ : accuracy of the first order growth index model @xmath124 . in all plots
we use the same color scale and we assume a flat cosmological model with @xmath125 . the relative error represents the maximal imprecision in the redshift range surveyed by a euclid - like survey ( @xmath126 ) . , title="fig : " ] parameters ( @xmath122 ) , in approximating the growth index of smooth dark energy models with the first order model @xmath123 of @xcite . _
middle panel _ : accuracy of the phenomenological fitting formula of @xcite .
_ right panel _ : accuracy of the first order growth index model @xmath124 . in all plots
we use the same color scale and we assume a flat cosmological model with @xmath125 . the relative error represents the maximal imprecision in the redshift range surveyed by a euclid - like survey ( @xmath126 ) . , title="fig : " ] parameters ( @xmath122 ) , in approximating the growth index of smooth dark energy models with the first order model @xmath123 of @xcite .
_ middle panel _ : accuracy of the phenomenological fitting formula of @xcite .
_ right panel _ : accuracy of the first order growth index model @xmath124 . in all plots
we use the same color scale and we assume a flat cosmological model with @xmath125 . the relative error represents the maximal imprecision in the redshift range surveyed by a euclid - like survey ( @xmath126 ) . , title="fig : " ] we first compute the coefficients @xmath127 ( see eq .
[ eq : gamma_coeff_w ] ) .
if we assume @xmath128 , then the limit @xmath129 is equivalent to the limit @xmath130 .
we obtain @xmath131_{a=0}\,,\ ; \mathcal{w}_1 = \big[({\omega}')^{-1 } w'\big]_{a=0 } \,,\ ; \mathcal{w}_2 = \big[({\omega}')^{-2}\big(3w'w-(w')^2/w+w''\big)-({\omega}')^{-1 } w'\big]_{a= 0 } \ , , \
] ] etc . , where @xmath132 from equation ( [ eq : relations_sg ] ) .
the formalism requires the knowledge of the @xmath133 order derivative of the de _ eos _
@xmath97 for @xmath134 , that is well beyond the redshift domain where the expansion rate of the universe is expected to be sensitive to contributions from the dark energy density .
noting that @xmath135 when @xmath134 , the finiteness of the coefficients @xmath136 ( at least up to @xmath137 ) can be enforced by requiring the _ eos _ to satisfy the following equation @xmath138 whose most general solution is @xmath139 where @xmath140 , @xmath141_{a=1}$ ] .
note that by linearizing the previous expression around @xmath142 one recovers the standard parameterization @xmath143 @xcite .
for this specific equation of state , we have @xmath144 and , as a consequence , the coefficients of the series expansion which defines the growth index are [ eq : gammanw(a)cdm ] @xmath145 the relative variation of the growth index @xmath0 with respect to the @xmath3cdm value , when the _ eos _ parameters span the range ( @xmath146 ) can be as large as @xmath147 at @xmath35 . as a reference ,
if @xmath148 is fixed to the value @xmath149 , then one can expect a relative difference as important as @xmath150 , a figure larger than the nominal precision with which a euclid - like survey is expected to constrain @xmath12 , that is @xmath151 @xcite .
figure [ fig : w(a)cdm_a ] shows that within the redshift range that will be surveyed by euclid ( @xmath126 ) , already when truncating the expansion at first order , the maximal imprecision with which the redshift scaling of @xmath12 is reconstructed over all the parameter space is less than 0.1% .
we are not aware of any theoretically justified model that accounts for possible variability in the de _ eos _ , so we compare the accuracy of our parameterization ( cf .
eqs . ) with that obtained using the wang & steinhardt ( 1998 ) model@xcite , which is supposed to be an accurate approximation also for weakly varying dark energy equation of states .
additionally , we also consider the phenomenological fitting formula of @xcite , ( @xmath152\,(1\,+\,w(z=1)\,)$ ] , where @xmath153 is the heaviside step function and @xmath108 is the constant equation of state ( _ eos _ ) parameter of the dark energy fluid ) . as can be appreciated by inspecting figure [ fig : w(a)cdm_a ] , the gain in precision with respect to both these growth index models can be as high as 1 order of magnitude in the range ( @xmath146 ) , a figure which pushes the systematic error well below the expected measurement precision of the euclid satellite . ) , in approximating the growth index of a clustering quintessence model @xcite with eq .
( 53 ) of @xcite ( _ left panel _ ) and with eq .
( [ eq : gamma_taylor ] ) truncated at first order ( _ right panel _ ) .
the relative error , calculated at @xmath154 , represents the maximal imprecision in the window surveyed by a euclid - like survey ( @xmath126).,title="fig : " ] ) , in approximating the growth index of a clustering quintessence model @xcite with eq .
( 53 ) of @xcite ( _ left panel _ ) and with eq .
( [ eq : gamma_taylor ] ) truncated at first order ( _ right panel _ ) .
the relative error , calculated at @xmath154 , represents the maximal imprecision in the window surveyed by a euclid - like survey ( @xmath126).,title="fig : " ] dark energy affects the process of structure formation not only through its equation of state , but also through its speed of sound .
indeed , if the speed of sound with which dark energy perturbations propagate drops below the speed of light , then dark energy inhomogeneities increase and couple gravitationally to matter perturbations . as a consequence , they may be detectable on correspondingly better observable ( though typically still large ) scales .
the influence of an unstable dark energy component on the clustering of matter can be effectively described by switching on the additional degrees of freedom @xmath49 and @xmath46 in eq .
( [ eq : matter_density_fluctuations ] ) . as an archetype of this class of phenomena , we consider the clustering dark energy model with vanishing speed of sound introduced in @xcite . in this model , the dark energy _
eos _ is assumed to be constant , the speed of sound is effectively null , and we have @xmath155 as a consequence , the structural parameters of our formalism are [ eq : cq_m_n ] @xmath156 while @xmath157 and @xmath158 are the same as in eq . .
using eq . , it follows immediately that the growth index coefficients are [ eq : claqui ] @xmath159 note that , as for smooth quintessence , the @xmath3cdm model is also included in clustering quintessence as the limiting case @xmath90 ( as a matter of fact , by setting @xmath90 in eqs .
we finde the coefficients ) . in figure
[ fig : sefuverni ] we compare the exact numerical calculation of the growth index @xmath0 and the approximation ( [ eq : gamma_taylor ] ) with coefficients @xmath30 and @xmath84 computed above .
the relative discrepancy , for the characteristic values of @xmath26 and @xmath108 is shown in figure [ fig : sefuverni ] . by comparing our parameterization of the growth index with that of sefusatti and vernizzi ( 2011 ) ( left panel of figure [ fig : sefuverni ] )
, we can appreciate the gain in precision , which is roughly a factor of 2 .
in this section we show how to predict the amplitude of the growth index in scenarios where the einstein fields equations are modified .
we will first discuss what has become a reference benchmark for alternative gravitational scenarios , the dgp model @xcite , and we will then generalize the discussion , by applying the formalism to post parametrized friedmann models @xcite . .
the precision is estimated as the relative difference with respect to the numerical reconstruction of the growth index @xmath0 in the reference model . ]
we consider the dvali - gabadadze - porrati brainworld model@xcite in which gravity is modified at large distances due to an extra spatial dimension .
the scale factor of the resulting cosmological model evolves according to the following modified friedmann equation @xcite @xmath160 where @xmath161 is the length scale at which gravity starts to leak out into the bulk . neglecting effects of spatial curvature ( @xmath86 ) and defining @xmath162 , the hubble rate can be expressed as @xmath163 \label{eq : dgp_hubble}\end{aligned}\ ] ] which implies the constraint @xmath164 . by deriving eq .
( [ eq : dgp_friedmann ] ) with respect to @xmath54 and by using the standard energy conservation equation @xmath165 , we find , in terms of the fundamental variable of our formalism ( @xmath67 ) @xmath166 from which it follows that the flat dgp model is formally equivalent to a de model with an @xmath87 varying as @xmath167 .
the effective newton constant in the poisson equation is @xcite @xmath168\,,\end{aligned}\ ] ] and , after some algebra , the source term @xmath169 in eq .
( [ eq : matter_density_fluctuations ] ) can be expressed as @xmath170 from eqs . , and we find the following structural parameters [ eq : dgp_x_h_m_n ] @xmath171 and , finally , the growth index coefficients @xmath172 in what follows we consider the flat dgp model with @xmath173 . with this choice of the density parameter ,
the dgp model best fits the expansion rate of the _ reference _ @xmath3cdm model in the range @xmath174 . in this non - standard gravity scenario , the maximal relative error , when comparing the parametric reconstruction of the growth rate with numerical results in the redshift range covered by euclid - like survey , is @xmath175 at order 1 , and @xmath176 at order 2 , see fig [ fig : comparison_dgp ] . to sub percentage precision we need to expand the growth index one order further .
the somewhat degraded precision for the dgp model is due to the fact that the taylor parameterization proposed in this work ( cf .
the approximation ) is specifically tailored for models in which the growth index minimally deviates from the @xmath3cdm prediction .
it is however impressive how the dgp model , despite its extreme nature , can still be satisfactorily described by our formalism .
we will now assume that , in viable theories of modified gravity , both background and perturbed observables have values close to that of the standard cosmological model .
that is , deviations from @xmath3cdm are hypothesized to be small , comparable with current observational uncertainties .
we further assume that eq .
( [ eq : matter_density_fluctuations ] ) is the master equation for studying matter collapse on sub - horizon size . in other terms ,
linear density fluctuations evolve in the quasi - static newtonian regime .
this being the case , the structure coefficients of the growth index formalism are given by eq .
( [ eq : struct ] ) and , by a straightforward implementation of formula ( [ eq : gamma_n ] ) , we obtain [ eq : mg_gamma_n ] @xmath177_{{\omega}=0 } \\ \gamma_{1 } & = & \bigg [ \frac{-(2 - 3\mu)\gamma_0 ^ 2 -6(\mu + \frac{d\mu}{d{\omega}}-w-2 \frac{dw}{d{\omega}})\gamma_0 - 3\mu + 6\frac{d\mu}{d{\omega } } + 3\frac{d^2\mu}{d{\omega}^2}-3w-6\frac{dw}{d{\omega}}- 2\frac{d^2\nu}{d{\omega}^2 } } { 2(2 + 3\mu - 12w ) } \bigg]_{{\omega}=0 } \\ \gamma_{2 } & = & \bigg[\frac{-(2 + 3\mu)\gamma_0 ^ 3 + 9(\mu+\frac{d\mu}{d{\omega}})\gamma_0 ^ 2 - 6(2 - 3\mu)\gamma_0\gamma_1 +
3\big(2w+6\frac{dw}{d{\omega}}+6\frac{d^2w}{d{\omega}^2}-3\mu-6\frac{d\mu}{d{\omega}}-3\frac{d^2\mu}{d{\omega}^2}\big)\gamma_0}{3(2 + 3\mu -18w ) } \nonumber \\ & & \;\ ; + \frac{18\big(2w+4\frac{dw}{d{\omega}}-\mu -\frac{d\mu}{d{\omega } } \big)\gamma_1 + 3\big(1+w+3\frac{dw}{d{\omega}}+3\frac{d^2w}{d{\omega}^2}+3\frac{d\mu}{d{\omega}}+3\frac{d^2\mu}{d{\omega}^2}+\frac{d^3\mu}{d{\omega}^3}\big)-2\frac{d^3\nu}{d{\omega}^3}}{3(2 + 3\mu - 18w ) } \bigg]_{{\omega}=0}\end{aligned}\ ] ] one can parameterize a large class of modified gravity scenarios by expanding the model dependent quantities @xmath49 and @xmath46 in power series .
since we are interested in alternative gravitational scenarios that might explain away the dark energy phenomenon , we expect deviations from einstein s gravity to become appreciable as @xmath178 starts to diverge from 0 . by assuming both @xmath49 and @xmath46 analytic at @xmath120 , we can thus expand ( see also @xcite ) [ eq : taylor_wmunu ] @xmath179 where we denoted by @xmath180 the value of the @xmath87 in the epoch of matter domination as in [ sec : w(a ) ] .
the corresponding growth index coefficients are [ eq : pmg_gamma_n ] @xmath181 , between the growth index predicted by the @xmath182cdm model of @xcite and the @xmath3cdm value , for various values of the parameters @xmath183 , and @xmath184 .
_ right panel : _ we plot the maximal relative error , in the redshift interval surveyed by euclid , of the first order approximation @xmath185 as a function of the amplitude of the parameters @xmath183 and @xmath184 specifying the @xmath182cdm model .
the imprecision is maximal at @xmath154 , and it is smaller than @xmath2 already when the parameterization given in eq .
[ eq : gamma_taylor ] is truncated at first order.,title="fig : " ] , between the growth index predicted by the @xmath182cdm model of @xcite and the @xmath3cdm value , for various values of the parameters @xmath183 , and @xmath184 .
_ right panel : _ we plot the maximal relative error , in the redshift interval surveyed by euclid , of the first order approximation @xmath185 as a function of the amplitude of the parameters @xmath183 and @xmath184 specifying the @xmath182cdm model .
the imprecision is maximal at @xmath154 , and it is smaller than @xmath2 already when the parameterization given in eq .
[ eq : gamma_taylor ] is truncated at first order.,title="fig : " ] as an example of the application of the formalism , we consider the @xmath182cdm scenario of @xcite .
this is a one - parameter family of models representing a large class of modified gravity theories for which the fundamental geometric degree of freedom is the metric , the field equations are at most 2nd order in the metric variables and gauge - invariant @xcite .
the interesting facet of these non - standard gravity models is that deviations from gr are parameterized by an observable , the _ gravitational slip _
parameter @xmath186 where @xmath187 and @xmath188 are the newton and curvature potentials respectively , both assumed to be time and scale dependent functions . in this non - standard gravity formalism , the background evolution is degenerate with that of the @xmath108cdm model of cosmology , that is it is effectively described by the friedmann equation ( ) augmented by a dark energy component @xmath189 with equation of state parameter @xmath108 , satisfying the usual conservation equation .
this fixes the amplitude of the structural parameters of our formalism to those of .
deviations from general relativity are expected only in the perturbed sector of the theory . at first order , indeed ,
this class of models predicts the following modification of the newton constant @xmath190 where it is assumed that on small cosmic scales , well inside the horizon , any scale dependence in the gravitational slip parameter can be neglected .
this in turns implies that the growth of matter perturbations , in these scenarios , can be described by inserting @xmath191 and @xmath192 into eq .
[ eq : f_h_general ] . since
, at early epochs deviations from general relativity are constrained by big bang nucleosynthesis measurements , it is a fair hypothesis to assume that @xmath193 is a smooth function which deviates from zero as soon as the dark energy comes to dominate the overall energy budget of the universe . as a result
, the gravitational slip parameter can be expanded as @xcite@xmath194 and the overall effects of non - standard gravity are governed by the set of discrete parameters @xmath195 . from eqs . and it is now easy to read off the value of the growth indices in the @xmath182cdm model : [ eq : ppf_gamma_n ] @xmath196 the amplitude of the non - standard signals expected in this alternative gravitational scenario is shown in the left panel of figure [ fig : ppf_variation ] where we display the distortions which a possibly non - null value of the @xmath182cdm parameters @xmath183 and @xmath184 induce on the growth index . also the accuracy with which the growth index is reconstructed by our formalism is shown ( left panel ) .
this last plot confirms that systematic uncertainties are below the threshold of the planck statistical errors over a region of the parameter space ( @xmath183 , @xmath184 ) which is sufficiently large to be physically interesting .
.compilation of currently available growth rate data . [ cols="<,<,^,^",options="header " , ]
besides being instrumental in increasing the accuracy with which the growth rate is reconstructed from data , the @xmath12-formalism introduced in the previous sections also serves as a guide in interpreting empirical results directly in terms of dark energy models .
this is shown in this section , where we confront the growth index predictions with current data and data simulations for a euclid - like survey . after describing our data analysis strategy
, we show here how we test whether the @xmath3cdm model correctly describes available data about the linear growth rate of structures and how we use the growth index parameter space @xmath197 to analyze and draw statistical conclusions on various non - standard gravity scenarios , in a manner which is independent from the specific details of the expansion history of the universe .
we first focus on the constraints that present day observations set on the growth indices @xmath1 .
these are derived by computing the likelihood @xmath198 of the data shown in table [ tab : data ] given the model in eqs . and .
to this purpose we minimize the @xmath199 function @xmath200 where @xmath201 is the uncertainty in the growth data , @xmath202 is the set of parameters that , except for @xmath203 , fix the background expansion , @xmath204 are the growth indices introduced in eq . and @xmath205 where @xmath206 is the growth factor . in this paper
we restrict our analysis to the first two growth indices , i.e. we set @xmath207 .
we also assume the @xmath208 measurements in table [ tab : data ] to be independent .
a more sophisticated error analysis might eventually results in minor changes in our quantitative conclusions , but will have no impact on their physical interpretation .
it is well known that @xmath208 can not be estimated from data without picking a particular model , or at least a parametrization , for gravity , i.e. , for the quantity being tested @xcite . despite the strong prior on the underlaying gravitational model , there is however evidence that the estimated values of this observable depend negligibly on the distance - redshift conversion model , that is for sensible variations of the background parameter @xmath209 in a flat @xmath3cdm model , the variation of the estimation is well below the statistical error @xcite .
this should guarantee that data can be meaningfully compared against models with distinct background evolution from that assumed to estimate the observable .
the likelihood analysis is carried out by choosing a prior model ( hereafter called _ fiducial _ ) for the evolution of @xmath31 and @xmath210 in eqs .
( [ eq : deff ] ) . to this purpose
we choose the _ reference _ @xmath3cdm model i.e. a flat @xmath3cdm model with @xmath4 parameters @xmath211 @xcite .
the resulting likelihood contours for @xmath30 and @xmath84 are shown in the left panel of figure [ fig : planck_wmap9 ] . by marginalizing in the growth index parameter space
we find that , at @xmath212 confidence level ( c.l . ) , the relative precision on the leading and first order growth indices is @xmath5 and @xmath6 . on the right panel of figure
[ fig : planck_wmap9 ] we show the same analysis assuming wmap9 background values @xmath213 . in this case , the 1-dimensional marginalized @xmath212 confidence levels are @xmath214 and @xmath215 .
the likelihood contours are appreciably smaller if the likelihood analysis is carried out in the wmap9 fiducial , owing to the fact that the statistical analysis depends on the background model .
we will see in the next section how to factor out the specific choice of the background model from the interpretation of growth rate data .
and @xmath14 confidence levels in the growth index parameter space @xmath216 ( orange filled regions ) obtained from the compilation of data shown in table [ tab : data ] and by using , as fiducial for the evolution of the background metric , the expansion rate of the _ reference _ @xmath3cdm model ( @xmath217 , @xmath218 ) .
the marginalized best - fitting values are @xmath5 and @xmath6 .
the growth indices theoretically expected in the fiducial @xmath3cdm model ( @xmath219 , @xmath8 ) are shown by the filled black square .
the black dotted line , defined by eq .
, shows the partition of the @xmath220 plane into regions where growth is amplified / suppressed with respect to the fiducial model . _
right panel : _ the same but using as fiducial the @xmath3cdm model with wmap9 values ( @xmath221 , @xmath222 ) .
the @xmath3cdm model is represented by the empty purple square.,title="fig : " ] and @xmath14 confidence levels in the growth index parameter space @xmath216 ( orange filled regions ) obtained from the compilation of data shown in table [ tab : data ] and by using , as fiducial for the evolution of the background metric , the expansion rate of the _ reference _ @xmath3cdm model ( @xmath217 , @xmath218 ) .
the marginalized best - fitting values are @xmath5 and @xmath6 .
the growth indices theoretically expected in the fiducial @xmath3cdm model ( @xmath219 , @xmath8 ) are shown by the filled black square .
the black dotted line , defined by eq . , shows the partition of the @xmath220 plane into regions where growth is amplified / suppressed with respect to the fiducial model . _ right panel : _ the same but using as fiducial the @xmath3cdm model with wmap9 values ( @xmath221 , @xmath222 ) .
the @xmath3cdm model is represented by the empty purple square.,title="fig : " ] the traditional way of exploiting the growth index formalism is to use it as a tool for a consistency test of the scenario proposed to model the background kinematics of the universe .
the goal is to check whether a given gravitational model that explains the observed cosmic expansion rate also predicts the correct growth rate of linear structures .
for example , one needs to verify that the most likely amplitude of the growth index parameters derived from observations of the linear growth of structures is not in conflict with those predicted on the basis of the de _ eos _ parameters @xmath148 and @xmath223 which best fit expansion rate data . in absence of major observational systematics ,
any possible mismatch between measured and expected values of the growth index , would be the smoking gun of new gravitational physics . in the opposite case ,
growth data provide additional constraints on dark energy parameters .
figure [ fig : planck_wmap9 ] shows that the growth index predicted on the basis of @xmath224 and @xmath225 , i.e. the _ eos _ values of the reference @xmath4 @xmath3cdm model agrees with results from the likelihood analysis of linear growth rate data coming from a variety of low redshift surveys of the large scale structure of the universe
. should nt this be the case , one could question either the unbiased nature of the data analysis in both sectors , background and perturbed , either the effectiveness of the standard description of gravity in terms of a @xmath3cdm framework .
a first immediate advantage of interpreting growth history data in the growth index plane , is that this parameter space facilitates the interpretation of critical information encoded in the likelihood function .
specifically , it allows to classify alternative dark energy models ( each labeled by the pair of coordinates @xmath226 ) as either generating more or less growth of structures with respect to the chosen fiducial .
the line separating these two characteristic regions is shown in figure [ fig : planck_wmap9 ] and it is computed by imposing @xmath227 where @xmath228 are the growth indices of the fiducial model ( that is @xmath229 in fig [ fig : planck_wmap9 ] ) , @xmath230 are the background parameters of the fiducial model and @xmath231 is the initial redshift at which perturbations are conventionally assumed start growing .
the region of more growth in the @xmath12-plane is defined as the region where a density perturbation ( whose amplitude is normalized to unity today ) had , at the initial redshift , a smaller amplitude than that predicted in the fiducial model ( @xmath232 ) whereas regions of weaker gravity are the @xmath12 loci where the amplitude of the perturbation was larger ( @xmath233 ) .
note that the orientation of the line separating these two regions depends only negligibly on the chosen initial redshift @xmath231 ( for the sake of illustration we have set @xmath234 in figure [ fig : planck_wmap9 ] ) .
our analysis shows that current data mildly favor models in which the strength of gravitational interactions is weaker than what is predicted in the _ reference _ @xmath3cdm model .
this result also confirms a conclusion of @xcite .
as compared to the way perturbations grow within purely matter - dominated models , it seems that data tend to prefer a suppression mechanism more efficient than that provided by the cosmological constant .
and @xmath14 confidence levels in the growth indices parameter space @xmath30 - @xmath84 ( orange filled regions ) for a euclid - like survey , assuming as fiducial cosmological model the _ reference _ @xmath3cdm model ( @xmath217 , @xmath218 ) .
the black dotted and blue solid lines show predictions of a smooth quintessence model , that is a dark energy component with variable _ eos _ of parameters @xmath148 and @xmath223 ( see eq .
[ eq : wi_finite ] ) .
these parameters span the range @xmath235 $ ] and @xmath236 $ ] respectively .
the spacing between adjacent lines is 0.05 and 0.2 respectively .
the growth indices theoretically expected in the fiducial @xmath3cdm model is shown by a filled black square . ]
once a fiducial for the background evolution is chosen to compute the likelihood function , what conclusions can we draw on the viability of gravitational models other than the fiducial itself ?
we will now see that , since the dependence of the growth index on the relevant cosmological parameters is explicitly taken into account in our analysis scheme , the likelihood in the @xmath220 plane , besides allowing to reject the null hypothesis that the fiducial is compatible with data , also allows to set constraints on alternative cosmological scenarios .
in other terms , it is possible to exploit model dependent likelihood contours in the @xmath197 plane to tell apart also those theoretical models characterized by an evolution of the background sector @xmath237 which is distinct from that of the fiducial model @xmath230 itself .
the growth indices @xmath30 and @xmath84 are model dependent quantities which can be estimated only once a specific background fiducial model for the evolution of @xmath31 is chosen ( cf .
. [ eq : deff0 ] ) . as a consequence ,
the growth indices inferred in distinct background models can not be directly compared .
for example , consider the likelihood contours in the plane @xmath197 obtained by assuming the _ reference _ @xmath3cdm model as fiducial ( figure [ fig : planck_wmap9 ] ) .
one can not constrain the de @xmath87 parameters @xmath148 and @xmath223 by simply computing the amplitude of the coefficients @xmath238 and @xmath239 expected in an effective de model with @xmath87 parameters @xmath148 and @xmath223 ( cf .
[ sec : w(a ) ] ) and by confronting these theoretical values with the empirical likelihood .
indeed , the specific growth rate history of a de model is not entirely captured by the growth indices , part of the information being locked in the scaling of the background density parameter @xmath31 .
we overcome this critical pitfall by computing @xmath240 , the amplitude of the effective growth indices in the fiducial background , that is the exponent that once inserted in ( [ eq : f_approximation ] ) together with the matter density parameter of the fiducial model adopted in the likelihood analysis ( @xmath241 ) , allows to match the scaling of the growth rate expected in the specific gravitational model under consideration .
this is equivalent to enforcing the following identity @xmath242 by means of this transformation strategy we factor out the effect of expansion from the analysis of growth rate histories .
we will illustrate this feature by simulating the constraints that the growth index parameters expected from a next generation survey such as the space mission euclid will put on the background dark energy parameters @xmath148 and @xmath223 .
the euclid mission is designed to survey , in spectroscopic mode , @xmath243 galaxies in the redshift range @xmath244 and in a sky area of @xmath245 deg@xmath246 @xcite .
we simulate the expected growth data @xmath247 assuming as fiducial , the reference @xmath3cdm model . to this purpose
, we simply split the redshift range @xmath126 into 14 intervals , and we predict @xmath248 by using eqs . , and .
we finally assume that the relative error on the observable , in each interval , is that corresponding to the euclid figure of merit listed in table 2.2 of @xcite , i.e. @xmath249 .
the resulting likelihood contours in the @xmath220 plane , obtained via a standard @xmath250 analysis , are shown in figure [ fig : euclid_wowa ] .
notice that the growth index figure of merit , defined as the inverse of the surface of the @xmath212 likelihood contour in the @xmath197 plane , is expected to increase by a factor @xmath251 when compared to that deduced from current constraints ( see figure [ fig : planck_wmap9 ] ) . in figure
[ fig : euclid_wowa ] we also show the effective growth indices ( @xmath252 predicted by various de models ( labeled by the @xmath87 parameters @xmath148 and @xmath223 ) obtained by means of the transformation law . in practice
we compute these effective growth indices ( @xmath253 ) by minimizing numerically the integral @xmath254 over the redshift range covered by observational data ( i.e. @xmath255 $ ] for current data and @xmath256 $ ] for euclid - data simulations ) .
we have verified that this mapping is sufficiently precise over all the redshift intervals where acceleration effects are observable .
for @xmath257 typical errors of order @xmath258 arise for the most extreme values of @xmath148 and @xmath223 shown in figure [ fig : euclid_wowa ] .
therefore , this error is largely negligible with respect to the precision of the constraints .
note that while varying @xmath148 and @xmath223 , we have kept fixed @xmath209 , that is we overlook any possible degeneracy between the de @xmath87 and matter density parameters entering the expansion rate @xmath259 .
this essentially because the relative variation induced in the distance modulus @xmath260 by this simplifying assumption is less than @xmath261 in the domain of interest .
overall , figure [ fig : euclid_wowa ] shows how measurements in the perturbed sector help to tighten the constraints on background cosmological parameters , and , ultimately to discriminate among de models .
specifically , we predict that growth rate data will provide independent constraints on the value of @xmath262 characterized by a relative(/absolute ) precision of @xmath263 . and @xmath14 confidence levels in the growth indices parameter space @xmath30 - @xmath84 ( orange filled regions ) for a euclid - like survey , assuming as fiducial of a flat clustering quintessence model @xcite with parameters @xmath217 and @xmath264 .
the theoretically predicted values @xmath265 for this fiducial model are shown by a light blue cross .
we also show the effective growth indices ( in the chosen fiducial ) for smooth and clustering quintessence models with constant @xmath87 parameter @xmath266 ( black and blue lines respectively ) .
the spacing between points is @xmath267 with @xmath108 values increasing along the direction specified by the arrow .
_ right panel : _ the same as the left panel but here we have overplotted the amplitude of the growth indices @xmath30 and @xmath84 predicted by the @xmath182cdm model of @xcite as a function @xmath268 $ ] ( red dashed isocontours ) and @xmath269 $ ] ( blue full isocontours ) .
the spacing between adjacent lines is @xmath270 and @xmath271 respectively .
the fiducial model is the reference @xmath3cdm model .
, title="fig : " ] and @xmath14 confidence levels in the growth indices parameter space @xmath30 - @xmath84 ( orange filled regions ) for a euclid - like survey , assuming as fiducial of a flat clustering quintessence model @xcite with parameters @xmath217 and @xmath264 .
the theoretically predicted values @xmath265 for this fiducial model are shown by a light blue cross .
we also show the effective growth indices ( in the chosen fiducial ) for smooth and clustering quintessence models with constant @xmath87 parameter @xmath266 ( black and blue lines respectively ) .
the spacing between points is @xmath267 with @xmath108 values increasing along the direction specified by the arrow .
_ right panel : _ the same as the left panel but here we have overplotted the amplitude of the growth indices @xmath30 and @xmath84 predicted by the @xmath182cdm model of @xcite as a function @xmath268 $ ] ( red dashed isocontours ) and @xmath269 $ ] ( blue full isocontours ) .
the spacing between adjacent lines is @xmath270 and @xmath271 respectively .
the fiducial model is the reference @xmath3cdm model .
, title="fig : " ] the remapping method illustrated by eq . provides a technique with which to tell apart gravitational models with identical background evolutions . in other terms
we can use the growth index parameter space to resolve the background degeneracy of two dark energy models , i.e. two models predicting the same background expansions , but different linear growth histories . to illustrate this feature , we consider two possible scenarios whose
background evolutions are degenerate with that of the @xmath272cdm model : the clustering quintessence and the @xmath182cdm models .
we first consider the case in which future euclid data are simulated in a clustering quintessence model , a cosmological model in which matter is supplemented by a quintessence component with null sound speed , i.e. the model of @xcite discussed in
[ sec : clustering ] . specifically , we forecast @xmath247 by using eqs . , and and by further imposing that the background expansion is effectively described in terms of the constant de _ eos _
parameter @xmath264 .
the likelihood contours in the @xmath220 plane , obtained by adopting as fiducial model the same clustering quintessence model used to simulate data , are displayed on the left panel of figure [ fig : same_background ] .
we now show how to use these measurements in the @xmath12 plane to tell apart the clustering quintessence from canonical smooth quintessence . to this purpose
we calculate the effective growth indices @xmath273 and @xmath274 in two different theoretical models of dark energy , the clustering quintessence and the smooth quintessence models ( both identified , for simplicity , via their constant @xmath87 parameter @xmath108 ) .
they are calculated by using eq . to map @xmath226 of eqs . and respectively into the effective values @xmath275 for the fiducial background @xmath276 used to analyze growth data . by comparing likelihood results vs. the predicted amplitude of the growth index
, we can appreciate how the smooth de model with the same background expansion as the _ true _ cosmological model ( in this context , the clustering quintessence model ) is clearly ruled out at 95% c.l . by growth rate data . specifically ,
if the effective _ eos _ deviates from the reference value @xmath90 by at least @xmath2 ( @xmath277 or @xmath278 ) a euclid - like survey has enough resolving power to discriminate between smooth and clustering quintessence . as an additional example , on the right panel of figure [ fig : same_background ]
we show the constraints that an euclid - like survey can set on the slip parameter entering the @xmath182cdm model reviewed in [ sec : skofe ] . to this purpose
we simulate euclid observations assuming the _ reference _ @xmath3cdm model and we then reconstruct the likelihood assuming this very same model as fiducial .
we then compare this statistic to prediction from the @xmath182cdm model , that is to the values @xmath30 and @xmath84 of eqs . computed for different values of the slip parameters @xmath183 and @xmath184 defined in equation ( for simplicity we here explore only models for which @xmath279 and @xmath280 ) .
note that , in this particular case , owing to the fact that the background evolution does not change from model to model , the remapping procedure is superfluous .
the right panel of figure [ fig : same_background ] displays the performances of an euclid - like survey in detecting possible deviations of the slip parameter @xmath182 from its gr value , _
i.e _ from zero . we can conclude that data will have enough resolving power to exclude , at the 95% c.l .
, models predicting a parameter @xmath183 larger than 0.025 , or in other terms data can detect a relative deviation between the newtonian and the curvature potential if it is larger than about 2.5% .
and @xmath14 confidence levels in the parameter plane @xmath216 obtained from the compilation of data shown in table [ tab : data ] and by using , as fiducial for the evolution of the background metric , the expansion rate of the _ reference _ @xmath3cdm model ( @xmath217 and @xmath218 ) .
the growth indices theoretically expected in the fiducial @xmath3cdm model is shown by the filled black square . the growth index expected in the flat dgp model ( black diamond ) which best fits the expansion rate of the _ reference _
@xmath3cdm model is obtained via , that is after mapping in the appropriate background model . _
right panel : _ the same as the left panel but using as fiducial the expansion rate of the dgp model @xcite instead.,title="fig : " ] and @xmath14 confidence levels in the parameter plane @xmath216 obtained from the compilation of data shown in table [ tab : data ] and by using , as fiducial for the evolution of the background metric , the expansion rate of the _ reference _ @xmath3cdm model ( @xmath217 and @xmath218 ) .
the growth indices theoretically expected in the fiducial @xmath3cdm model is shown by the filled black square . the growth index expected in the flat dgp model ( black diamond ) which best fits the expansion rate of the _ reference _
@xmath3cdm model is obtained via , that is after mapping in the appropriate background model . _
right panel : _ the same as the left panel but using as fiducial the expansion rate of the dgp model @xcite instead.,title="fig : " ] , but now , in addition , we show predictions for different values of @xmath209 ( violet line ) , the constant @xmath87 for the smooth de ( sde ) model ( black line ) and the clustering de ( cde ) model ( light blue line ) . the parameter ranges are @xmath281 , @xmath282 ( sde ) and @xmath283 ( cde ) .
the spacing is 0.05 for @xmath209 , 0.1 for sde and 0.05 for cde .
the prediction for the wmap9 @xmath3cdm model are shown by a violet empty square and the black triangle corresponds to a @xmath182cdm model with ( @xmath90 , @xmath284 , @xmath285 ) . ] up to this point , we have shown that models with distinct background evolutions and models with distinct growth rate predictions can be analyzed in the same parameter space , thanks to the remapping scheme of eq . .
a neat way to demonstrate the precision and consistency of this strategy is by showing that the conclusions on the physical viability of a model are invariant with respect to the choice of the fiducial in which data are analyzed .
we will demonstrate this key feature by exploiting the dgp@xcite model which predicts both different background and different growth rate evolutions with respect to the @xmath3cdm model . to this purpose , as in
[ sec : dgp ] we consider the flat dgp model which best fits the expansion rate of the _ reference _ @xmath3cdm model .
this dgp model has parameters @xmath286 , where the effective dgp @xmath87 is given in
[ sec : dgp ] . in figure
[ fig : lcdm_dgp ] we show the constraints set on the dgp model by current data ( cf .
table [ tab : data ] ) . in the left panel ,
we show the likelihood contours in the parameter plane @xmath220 obtained by using as fiducial for the evolution of the background metric the expansion rate of the _ reference _ @xmath3cdm model . the prediction for the dgp model obtained with the mapping is also shown . in the right panel , instead , we show the likelihood contours in the parameter plane @xmath220 obtained by using the expansion rate of the dgp model as fiducial for the evolution of the background metric . in this case
, it is the predictions of the @xmath3cdm model that are remapped via eq . .
by confronting these two plots we see that while the location , the amplitude and the orientation of the likelihood surfaces do vary , the interpretation of the results is clearly unchanged .
both figures show , consistently , that measurements of the growth rate in the range @xmath287 have already a sufficient precision to rule out , at a @xmath14 c.l .
, a most extreme scenario of modified gravity such as the dgp model . in other terms , the structure of the likelihood contours depends on the specific fiducial model of the background , that is each fiducial background defines its own growth index parameter space . but physical interpretation is unchanged if we change from one parameter space to the other using the transformation equation .
the advantage of the method is also illustrated in figure [ fig : all ] using current data . in this picture
we simultaneously contrast predictions from a large class of cosmological models against the likelihood contours derived assuming the _ reference _ @xmath3cdm as fiducial for data analysis . on top of the dgp model ( black diamond ) , we also plot the effective growth indices associated to the @xmath3cdm model which best fits wmap9 data @xcite ( purple empty square ) . in this scenario , characterized by the reduced density parameter @xmath221
, the growth of structures is slightly suppressed with respect to the fiducial , i.e. the _ reference _ @xmath3cdm model of planck .
one might also remark the slight ( but statistically insignificant ) tendency of growth data to favor the wmap9 substantiation of the @xmath3cdm model with respect to the planck one .
this is a particular example of a general feature of @xmath3cdm models : the higher the matter density and the more severe the tension with growth index constraints . for example
@xmath3cdm models with @xmath288 are excluded at @xmath14 confidence level . in figure
[ fig : all ] we also show the constraints on de models with constant @xmath87 parameter @xmath108 .
smooth de models with @xmath289 and clustering de models with @xmath290 are excluded at @xmath14 confidence level . finally , as a curiosity , we show that a @xmath182cdm model ( black empty triangle ) with the same background as the _ reference _ @xmath3cdm model but with a slip parameter @xmath284 best fits observations , that is it maximizes the likelihood of current data .
the observational information about the growth rate history @xmath13 of linear cosmic structures can be satisfactorily encoded into a small set of parameters , the growth indices @xmath1 , whose amplitude can be analytically predicted by theory .
their measurement allows to explore whether einstein s field equations encompass gravity also in the infrared , i.e. on very large cosmic scales . in order for this to be accomplished ,
_ a _ ) an optimal scheme for compressing the growth rate function into the smallest possible set of discrete scalars @xmath1 , without sacrifying accuracy , and _
b _ ) a prescription for routinely calculating their amplitude in relevant theories of gravity , in order to explore the largest region in the space of all possible models , must be devised . in this paper
we have explored a promising approach towards this goal .
we have demonstrated both the precision and the flexibility of a specific parameterization of the growth index , that is the logarithmic expansion .
if the fiducial gravitational model is not too different from standard gr , i.e. possible deviations in both the background and perturbed sector can be interpreted as first order correction to the friedmann model , then the proposed parameterization scheme allows to match numerical results on the redshift dependence of the growth index with a relative error which is lower than the nominal precision with which the next generation of redshift surveys are expected to fix the scaling of this function .
the performances are demonstrated by comparing , for various fiducial gravitational models , the accuracy of our proposal against that of different parameterizations available in the literature . besides accuracy , the formalism features two other critical merits , one practical and one conceptual .
first we supply a simple way for routinely calculating the amplitude of the growth indices in any gravitational model in which the master equation for the growth of density perturbations reduces to the form of eq .
[ eq : matter_density_fluctuations ] . to this purpose
it is enough to specify three characteristic functions of this equation , the expansion rate @xmath11 , the damping @xmath291 and the response @xmath292 coefficients to calculate the parameters @xmath1 up to any desired order @xmath293 .
moreover , since the parameterization of the growth rate has not a phenomenological nature , but it is constructed as a series expansion of the exact solutions of the differential equation which rules the growth of structures ( cf .
[ eq : f_h_general ] ) , one can easily interpret empirical results about the amplitude of the growth indices in terms of fundamental gravitational models .
since the growth index is a model dependent quantity , it has been traditionally used only to reject , statistically , the specific model adopted to analyze growth data .
we have shown , instead , that the growth index parameter space @xmath197 provides a diagnostic tool to discriminate a large class of models , even those presenting background evolution histories different from the fiducial model adopted in data analysis .
in other terms , a detection of a present day growth index amplitude @xmath294 would not only indicate a deviation from @xmath3cdm predictions but could be used to disentangle among different alternative explanations of the cosmic acceleration in a straightforward way .
the key to this feature is the mapping of eq .
[ eq : f_approximation2 ] which allows to factor out the effect of expansion from the analysis of growth rate histories .
as the standard @xmath295 plane identifies different expansion histories @xmath11 , the @xmath197 plane can thus be used to locate different growth rate histories @xmath13 .
we have illustrated the performance of the growth index plane in relation to modify gravity model selection / exclusion by using current data as well as forecasts from future experiments .
we have shown that the likelihood contours in the growth index plane @xmath220 can be used to tell apart a clustering quintessence component @xcite from a smooth dark energy fluid , to fix the parameters of viable parameterized post friedman gravitational models @xcite or to exclude specific gravitational models such as , for example , dgp @xcite .
the performances of the analysis tool presented in this paper are expected to be enhanced , should the formalism be coupled to models parameterizing the large class of possible gravitational alternatives to standard gr available in the literature . in particular various approaches
have been proposed to synthetically describe all the possible gravitational laws generated by adding a single scalar degree of freedom to einstein s equations @xcite .
besides quintessence , scalar - tensor theory and @xmath296 gravity , this formalism allows also to describe covariant galileons @xcite , kinetic gravity braiding @xcite and horndeski / generalized galileons theories @xcite .
interestingly , the cosmological perturbation theory of this general class of models can be parameterized so that a direct correspondence between the parameterization and the underlying space of theories is maintained . in a different paper
@xcite we have already explored how the effective field theory formalism of @xcite allows to interpret the empirical constraints on @xmath1 directly in terms of fundamental gravity theories .
we acknowledge useful discussions with l. guzzo , f. piazza , t. schcker , p. taxil and j. m. virey .
cm is grateful for support from specific project funding of the _ institut universitaire de france _ and of the labex ocevu . 99 s. perlmutter
_ et al . _
( sncp collaboration ) , `` measurements of @xmath10 and @xmath3 from 42 high - redshift supernovae '' , apj * 517 * , 565 ( 1999 ) [ astro - ph/9812133 ] a. g. riess _
et al . _
( snst collaboration ) , `` observational evidence from supernovae for an accelerating universe and a cosmological constant '' , aj * 116 * , 1009 ( 1998 ) [ astro - ph/9805201 ] p. astier _
et al . _
( snls collaboration ) , `` the supernova legacy survey : measurement of @xmath297 , @xmath298 and @xmath108 from the first year data set '' , a&a * 447 * , 31 ( 2006 ) [ astro - ph/0510447 ] c. marinoni and a. buzzi , `` a geometric measure of dark energy with pairs of galaxies '' , nature * 468 * , 539 ( 2010 ) e. komatsu _ et al . _ ( wmap collaboration ) , `` seven - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological interpretation '' , apjs * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 ] g. hinshaw _ et al . _
( wmap collaboration ) , `` nine - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological parameter results '' , apjs * 208 * , 19 ( 2013 ) [ arxiv:1212.5226 ] p. a. r. ade _ et al . _
( @xmath4 collaboration ) , `` planck 2013 results .
cosmological parameters '' , [ arxiv:1303.5076 ] l. anderson , e. aubourg , s. bailey , _ et al . _ ,
`` the clustering of galaxies in the sdss - iii baryon oscillation spectroscopic survey : baryon acoustic oscillations in the data release 9 spectroscopic galaxy sample '' , mnras * 427 * , 3435 ( 2013 ) [ arxiv:1203.6594 ] j. bel , c. marinoni , b. r. granett , l. guzzo , j. a. peacock , e. branchini _ et al .
_ , `` the vimos public extragalactic redshift survey ( vipers ) : @xmath299 from the galaxy clustering ratio measured at @xmath300 '' , a&a * 563 * , a37 ( 2014 ) [ arxiv:1310.3380 ] a. g. sanchez , c. g. scccola , a. j. ross _
et al . _ ,
`` the clustering of galaxies in the sdss - iii baryon oscillation spectroscopic survey : cosmological implications of the large - scale two - point correlation function '' , mnras * 425 * , 415 ( 2012 ) [ arxiv:1203.6616 ] p. j. e. peebles and b. ratra , `` the cosmological constant and dark energy '' , rev .
phys . * 75 * , 559 ( 2003 ) [ astro - ph/0207347 ] l. amendola and s. tsujikawa , `` dark energy : theory and observations '' , cambridge u. p. ( 2011 ) e. j. copeland , m. sami and s. tsujikawa , `` dynamics of dark energy '' , int .
j. mod .
d * 15 * , 1753 ( 2006 ) [ hep - th/0603057 ] c. wetterich , `` cosmology and the fate of dilatation symmetry '' , nucl . phys .
b * 302 * 668 ( 1988 ) r. r. caldwell , r. dave and p. j. steinhardt , `` cosmological imprint of an energy component with general equation of state '' , phys .
lett . * 80 * , 1582 ( 1998 ) [ astro - ph/9708069 ] c. armendariz - picon , v. f. mukhanov and p. j. steinhardt , `` a dynamical solution to the problem of a small cosmological constant and late - time cosmic acceleration '' , phys .
lett . * 85 * , 4438 ( 2000 ) [ astro - ph/0004134 ] p. binetruy , `` cosmological constant versus quintessence '' , int .
. phys . * 39 * , 1859 ( 2000 ) [ hep - ph/0005037 ] j. p. uzan , `` cosmological scaling solutions of nonminimally coupled scalar fields '' , phys .
d * 59 * ( 1999 ) 123510 [ gr - qc/9903004 ] a. riazuelo and j. p. uzan , `` cosmological observations in scalar - tensor quintessence '' , phys .
d * 66 * 023525 ( 2002 ) [ astro - ph/0107386 ] m. gasperini , f. piazza and g. veneziano , `` quintessence as a runaway dilaton '' , phys .
d * 65 * , 023508 ( 2002 ) [ gr - qc/0108016 ] t. clifton , p. g. ferreira , a. padilla and c. skordis , `` modified gravity and cosmology '' , phys . rept .
* 513 * , 1 ( 2012 ) [ arxiv:1106.2476 ] a. de felice and s. tsujikawa , `` @xmath296 theories '' , living rev .
* 13 * , 3 ( 2010 ) [ arxiv:1002.4928 ] g. r. dvali , g. gabadadze , and m. porrati , `` 4-d gravity on a brane in 5-d minkowski space '' , phys .
b. * 485 * , 208 ( 2000 ) [ hep - th/0005016 ] c. deffayet , g. r. dvali and g. gabadadze , `` accelerated universe from gravity leaking to extra dimensions '' , phys .
d * 65 * , 044023 ( 2002 ) [ astro - ph/0105068 ] n. arkani - hamed , s. dimopoulos , g. dvali and g. gabadadze , `` nonlocal modification of gravity and the cosmological constant problem '' [ arxiv : hep - th/0209227 ] s. capozziello , s. carloni and a. troisi , `` quintessence without scalar fields '' , recent res .
astrophys .
* 1 * , 625 ( 2003 ) [ arxiv : astro - ph/0303041 ] s. i .
nojiri and s. d. odintsov , `` modified gauss - bonnet theory as gravitational alternative for dark energy '' , phys .
b * 631 * , 1 ( 2005 ) [ hep - th/0508049 ] c. de rham , g. gabadadze and a. j. tolley , `` resummation of massive gravity '' , phys .
lett . * 106 * , 231101 ( 2011 ) [ arxiv:1011.1232 ] f. piazza , `` the ir - completion of gravity : what happens at hubble scales ? '' , new j. phys . *
11 * , 113050 ( 2009 ) [ arxiv:0907.0765 ] g. gubitosi , f. piazza and f. vernizzi , `` the effective field theory of dark energy '' , jcap * 1302 * , 032 ( 2013 ) [ arxiv:1210.0201 ] j. gleyzes , d. langlois , f. piazza and f. vernizzi , `` essential building blocks of dark energy '' , jcap * 1308 * , 025 ( 2013 ) [ arxiv:1304.4840 ] t. baker , p. g. ferreira and c. skordis , `` the parameterized post - friedmann framework for theories of modified gravity : concepts , formalism and examples '' , phys .
d * 87 * , 024015 ( 2013 ) [ arxiv:1209.2117 ] j. bel and c. marinoni , `` determination of the abundance of cosmic matter via the cell count moments of the galaxy distribution '' , a&a * 563 * , a36 ( 2014 ) [ arxiv:1210.2365 ] s. j. turnbull , m. j. hudson , h. a. feldman _
et al . _ ,
`` cosmic flows in the nearby universe from type ia supernovae '' , mnras * 420 * , 447 ( 2012 ) [ arxiv:1111.0631v2 ] m. davis , a. nusser , k. l. masters _
et al . _ , `` local gravity versus local velocity : solutions for @xmath301 and nonlinear bias '' , mnras * 413 * , 2906 ( 2011 ) [ arxiv:1011.3114 ] f. beutler , c. blake , m. colless _ et al .
_ , `` the 6df galaxy survey : @xmath302 measurement of the growth rate and @xmath210 '' , mnras * 423 * , 3430 ( 2012 ) [ arxiv:1204.4725 ] w. j. percival _ et al . _ ( 2dfgrs coll . ) , `` the 2df galaxy redshift survey : spherical harmonics analysis of fluctuations in the final catalogue '' , mnras * 353 * , 1201 ( 2004 ) [ astro - ph/0406513 ] y .- s .
song and w. j. percival , `` reconstructing the history of structure formation using redshift distortions '' , jcap * 10 * , 4 ( 2009 ) [ arxiv:0807.0810 ] n. p. ross , j. da ngela , t. shanks _
et al . _ , `` the 2df - sdss lrg and qso survey : the 2-point correlation function and redshift - space distortions '' , mnras * 381 * , 573 ( 2007 ) [ astro - ph/0612400 ] a. cabr and e. gaztaaga , `` clustering of luminous red galaxies i : large scale redshift space distortions '' , mnras * 393 * , 1183 ( 2009 ) [ arxiv:0807.2460 ] l. samushia , w. j. percival , a. raccanelli , `` interpreting large - scale redshift - space distortion measurements '' , mnras * 420 * , 2102 ( 2012 ) [ arxiv:1102.1014 ] b. a. reid , l. samushia , m. white _ et al . _ ,
`` the clustering of galaxies in the sdss - iii baryon oscillation spectroscopic survey : measurements of the growth of structure and expansion rate at @xmath303 from anisotropic clustering '' [ arxiv:1203.6641 ] c. contreras _ et al . _ ( wigglez collaboration ) , `` the wigglez dark energy survey : measuring the cosmic growth rate with the two - point galaxy correlation function '' , mnras advance access ( 2013 ) [ arxiv:1302.5178 ] l. guzzo , m. pierleoni , b. meneux _ et al .
_ , `` a test of the nature of cosmic acceleration using galaxy redshift distortions '' , nature * 451 * , 541 ( 2008 ) [ arxiv:0802.1944 ] s. de la torre , l. guzzo , j. peacock _ et al . _ , `` the vimos public extragalactic redshift survey ( vipers ) .
galaxy clustering and redshift - space distortions at @xmath304 in the first data release '' [ arxiv:1303.2622 ] l. samushia , b. a. reid , m. white _ et al .
_ , `` the clustering of galaxies in the sdss - iii baryon oscillation spectroscopic survey : measuring growth rate and geometry with anisotropic clustering '' [ arxiv:1312.4899 ] r. laureijs _ et al .
_ ( euclid consortium ) , `` euclid definition study report '' [ arxiv:1110.3193 ] p. j. e. peebles , `` the large scale structure of the universe '' , princeton university press ( 1980 ) l. wang , p. j. steinhardt , `` cluster abundance constraints on quintessence models '' , apj * * 508 * * , 483 ( 1998 ) [ astro - ph/9804015 ] e. v. linder , `` cosmic growth history and expansion history '' , phys .
d * 72 * , 043529 ( 2005 ) [ astro - ph/0507263 ] y. gong , `` the growth factor parameterization and modified gravity '' , phys .
d * 78 * , 123010 ( 2008 ) [ arxiv:0808.1316 ] r. gannouji , b. moraes , and d. polarski , `` the growth of matter perturbations in @xmath296 models '' , jcap * 0902 * , 034 ( 2009 ) [ arxiv:0809.3374 ] p.
wu , h. yu , x. fu , `` a parametrization for the growth index of linear matter perturbations '' , jcap * 0906 * , 019 ( 2009 ) [ arxiv:0905.3444 ] d. polarski and r. gannouji , `` on the growth of linear perturbations '' , phys .
b * 660 * , 439 ( 2008 ) [ arxiv:0710.1510 ] r. gannouji and d. polarski , `` the growth of matter perturbations in some scalar - tensor de models '' , jcap * 05 * , 018 ( 2008 ) [ arxiv:0802.4196 ] p. g. ferreira , c. skordis , `` the linear growth rate of structure in parametrized post friedmannian universes '' , phys .
d * 81 * , 104020 ( 2010 ) [ arxiv:1003.4231 ] m. kunz , d. sapone , `` dark energy versus modified gravity '' , phys .
lett . * 98 * , 121301 ( 2007 ) [ astro - ph/0612452 ] v. silvera and i. waga , `` decaying lambda cosmologies and power spectrum '' , phys .
d * 50 * , 4890 ( 1994 ) m. sereno and j. a. peacock , `` imprints of deviations from the gravitational inverse - square law on the power spectrum of mass fluctuations '' , mnras * 371 * , 719 ( 2006 ) [ astro - ph/0605498 ] l. knox , y .- s .
song , and j. a. tyson , `` distance - redshift and growth - redshift relations as two windows on acceleration and gravitation : dark energy or new gravity ? '' , phys .
d * 74 * , 023512 ( 2006 ) [ astro - ph/0503644 ] m. ishak , a. upadhye , and d. n. spergel , `` probing cosmic acceleration beyond the equation of state : distinguishing between dark energy and modified gravity models '' , phys .
d * 74 * , 043513 ( 2006 ) [ astro - ph/0507184 ] d. sapone and l. amendola , `` constraining the growth factor with baryon oscillations '' [ arxiv:0709.2792 ] s. nesseris and l. perivolaropoulos , `` testing @xmath3cdm with the growth function @xmath305 : current constraints '' , phys
d * 77 * , 023504 ( 2008 ) [ arxiv:0710.1092 ] c. di porto and l. amendola , `` observational constraints on the linear fluctuation growth rate '' , phys .
d * 77 * , 083508 ( 2008 ) [ arxiv:0707.2686 ] a. bueno belloso , j. garcia - bellido and d. sapone , `` a parametrization of the growth index of matter perturbations in various dark energy models and observational prospects using a euclid - like survey '' [ arxiv:1105.4825 ] c. di porto , l. amendola and e. branchini , `` growth factor and galaxy bias from future redshift surveys : a study on parametrizations '' , mnras * 419 * , 985 ( 2012 ) [ arxiv:1101.2453 ] v. cardone , s. camera and a. diaferio , `` an updated analysis of two classes of f(r ) theories of gravity '' , jcap * 1202 * , 030 ( 2012 ) [ arxiv:1201.3272 ] s. basilakos and a. pouri , `` the growth index of matter perturbations and modified gravity '' , mnras * 423 * , 3761 ( 2012 ) [ arxiv:1203.6724 ] f. piazza , h. steigerwald and c. marinoni , `` phenomenology of dark energy : exploring the space of theories with future redshift surveys '' ( 2013 ) [ arxiv:1312.6111 ] p. creminelli , g. damico , j. norena and f. vernizzi , `` the effective theory of quintessence : the @xmath107 side unveiled '' , jcap * 0902 * , 018 ( 2009 ) [ arxiv:0811.0827 ] m. chevallier and d. polarski , `` accelerating universes with scaling dark matter '' , int . j. mod .
d * 10 * , 213 ( 2000 ) [ arxiv : gr - qc/0009008 ] m. motta , i. sawicki , i. d. saltas , l. amendola , and m. kunz , `` probing dark energy through scale dependence '' , phys .
d * 88 * , 124035 ( 2013 ) [ arxiv:1305.0008 ] a. lue , `` the phenomenology of dvali - gabadadze - porrati cosmologies '' , phys .
* 423 * , 1 ( 2006 ) [ astro - ph/0510068 ] e. sefusatti and f. vernizzi , `` cosmological structure formation with clustering quintessence '' , jcap * 1103 * , 047 ( 2011 ) [ arxiv:1101.1026 ] c. skordis , `` consistent cosmological modifications to the einstein equations '' , phys .
d * 79 * , 123527 ( 2009 ) [ arxiv:0806.1238 ] j. k. bloomfield , e. e. flanagan , m. park and s. watson , `` dark energy or modified gravity ? an effective field theory approach '' , jcap * 08 * , 010 ( 2013 ) [ arxiv:1211.7054 ] a. nicolis , r. rattazzi and e. trincherini , `` the galileon as a local modification of gravity '' , phys .
d * 79 * , 064036 ( 2009 ) [ arxiv:0811.2197 ] c. deffayet , o. pujolas , i. sawicki and a. vikman , `` imperfect dark energy from kinetic gravity braiding '' , jcap * 1010 * , 026 ( 2010 ) [ arxiv:1008.0048 ] g. w. horndeski , `` second - order scalar - tensor field equations in a four - dimensional space '' , int .
. phys . * 10 * , 363 ( 1974 ) c. deffayet , g. esposito - farse and a. vikman , `` covariant galileon '' , phys .
d * 79 * , 084003 ( 2009 ) [ arxiv:0901.1314 ]
we compute the amplitude of the growth indices as follows .
first we introduce eq . into eq . and obtain @xmath306 neglecting the term on the right hand side , we have transformed the first order non - linear differential equation ( [ eq : f_h_general ] ) into an algebraic equation of the form @xmath307 now we set @xmath308 in eq .
and we obtain @xmath309 we suppose that @xmath310 , @xmath311 and @xmath312 in any viable model . in order to compute the coefficients
@xmath71 up to an order @xmath72 , we have to expand @xmath313 as a taylor series of @xmath314 at @xmath308 up to an order @xmath315 : @xmath316_{{\omega}=0 } + \mathcal{o}({\omega}^{n\:\!\ ! + \:\!\ ! 1 } ) .
\label{eq : f_taylor}\end{aligned}\ ] ] eq .
is satified if and only if all the coefficients vanish : @xmath317_{{\omega}=0 } = 0 \ ; , \qquad m=1,2,3,\ldots , n+1 \label{eq : df_0}\end{aligned}\ ] ] this leads naturally to the definition of the structural coefficients . for @xmath318 in eq .
we obtain eq . .
finally , inverting eq
. for @xmath319 leads to the recursion formula . | measurements of the growth index of linear matter density fluctuations @xmath0 provide a clue as to whether einstein s field equations encompass gravity also on large cosmic scales , those where the expansion of the universe accelerates .
we show that the information encoded in this function can be satisfactorily parameterized using a small set of coefficients @xmath1 , in such a way that the true scaling of the growth index is recovered to better than @xmath2 in most dark energy and dark gravity models .
we find that the likelihood of current data , given this formalism and the @xmath3 cold dark matter ( @xmath3cdm ) expansion model of @xmath4 , is maximal for @xmath5 and @xmath6 , a measurement compatible with the @xmath3cdm predictions ( @xmath7 , @xmath8 ) .
in addition , data tend to favor models predicting slightly less growth of structures than the @xmath4 @xmath9cdm scenario .
the main aim of the paper is to provide a prescription for routinely calculating , in an analytic way , the amplitude of the growth indices @xmath1 in relevant cosmological scenarios , and to show that these parameters naturally define a space where predictions of alternative theories of gravity can be compared against growth data in a manner which is independent from the expansion history of the cosmological background . as the standard @xmath10-plane provides a tool to identify different expansion histories @xmath11 and their relation to various cosmological models ,
the @xmath12-plane can thus be used to locate different growth rate histories @xmath13 and their relation to alternatives model of gravity . as a result
, we find that the dvali - gabadadze - porrati gravity model is rejected with a @xmath14 confidence level . by simulating future data sets , such as those that a euclid - like mission will provide ,
we also show how to tell apart @xmath3cdm predictions from those of more extreme possibilities , such as smooth dark energy models , clustering quintessence or parameterized post - friedmann cosmological models .
+ + |
let @xmath6 be an odd prime , and @xmath5 be an algebraically closed field of characteristic @xmath6 . for @xmath7 an integer , we consider the symplectic group @xmath4 , thought of as an algebraic group of rank @xmath8 .
it is well - known that the classification ( due to chevalley ) of rational simple @xmath4-modules is the same as in characteristic zero ( see jantzen ( * ? ? ?
* ii.2 ) ) .
more precisely , for every dominant weight @xmath9 there is a simple module @xmath3 , and these exhaust all isomorphism classes of simple modules . here
the set of dominant weights is the same as in characteristic zero : @xmath9 is dominant iff it is a linear combination of the fundamental weights @xmath10 ( @xmath11 ) with nonnegative integer coefficients . on the other hand , while the dimensions of simple @xmath12-modules can be computed from the weyl character formula , it seems that explicit dimension formulae for the modules @xmath13 for @xmath14 are quite rare , except in rather special situations .
we refer to @xcite for a survey . for fundamental weights , premet and suprunenko @xcite gave an algorithm to compute the dimensions of @xmath15 by reducing the problem to known properties of symmetric group representations .
later , gow @xcite gave an explicit construction of @xmath15 for the last @xmath2 fundamental weights ( that is : @xmath10 where @xmath16 ) which allowed him to obtain a recursive formula for their dimensions .
even later , foulle @xcite obtained a dimension formula for all fundamental weights .
as for other weights , it is known that for weights @xmath9 in the fundamental alcove the dimension of @xmath3 is the same as the dimension of @xmath17 ( the corresponding simple module in characteristic zero ) , and can thus be computed by the weyl character formula .
but for weights outside the fundamental alcove , no general dimension formula is known .
a conjectural formula by lusztig for primes in a certain range was shown to hold for @xmath18 by andersen - jantzen - soergel @xcite but was recently shown not to hold for all @xmath6 in the hoped - for range by williamson @xcite . in this paper , we show that topological quantum field theory ( tqft ) can give new information about the dimensions of some of these simple modules .
specifically , we show that for every prime @xmath0 and in every rank @xmath1 , there is a family of @xmath2 dominant weights @xmath19 , lying outside of the fundamental alcove except for one weight in rank @xmath20 , for which we can express the dimension of @xmath13 by formulae similar to the verlinde formula in tqft .
we found this family as a byproduct of integral @xmath21-tqft @xcite , an integral refinement of the witten - reshetikhin - turaev tqft associated to @xmath21 .
more precisely , we use integral @xmath21-tqft in what we call the ` equal characteristic case ' which we studied in @xcite . the family of weights @xmath19 we found together with our formulae for @xmath22 is given in the following theorem [ 1.1 ] . we can also compute the weight space decomposition of @xmath13 for these weights @xmath19 ; this will be given in theorem [ 1.8 ] .
we follow the notation of ( * ? ? ?
* planche iii ) , where the fundamental weights @xmath23 are expressed in the usual basis @xmath24 ( @xmath25 ) of weights of the maximal torus as @xmath26 .
let @xmath0 be prime and put @xmath27 . for rank @xmath1 ,
consider the following @xmath2 dominant weights for the symplectic group @xmath4 : @xmath28 put @xmath29 in case @xmath30 and @xmath31 and @xmath32 in case @xmath33 and @xmath34 .
then @xmath35 @xmath36 @xmath37 and @xmath38 is the same @xmath38 used in the definition of @xmath9 , except in case @xmath30 and @xmath34 , where we put @xmath39 . in case @xmath40 in rank @xmath20 , @xmath41 should be interpreted as zero .
formula ( [ verl ] ) is an instance of the famous verlinde formula in tqft .
formula ( [ d - verl ] ) appeared first in @xcite .
note that the difference between the two formulae is that certain sines in ( [ verl ] ) have become cosines in ( [ d - verl ] ) , and the overall prefactor is different . for fixed @xmath8 ,
both @xmath42 and @xmath43 can be expressed as polynomials in @xmath6 and @xmath38 . see @xcite for more information and further references . in appendix [ app - b ] , we give explicit polynomial expressions for the dimensions of our @xmath13 in rank @xmath44 .
when @xmath45 , the list above produces ( in order ) the fundamental weights @xmath46 .
these are exactly the weights considered by gow @xcite . for @xmath47 ,
our weights are different from those of gow .
it is intriguing that both gow s and our family of weights have @xmath2 elements
. can one find similar verlinde - like dimension formulae for other families of dominant weights ? in @xcite , we answered this question affirmatively for the @xmath2 fundamental weights considered by gow .
but @xcite was based on gow s recursion formula , not on tqft as in the present paper . on the other hand , @xmath21-tqft is just one of the simplest tqfts within the family of witten - reshetikhin - turaev tqfts , and it is conceivable that other integral tqfts ( _ e.g. _ @xcite ) might produce more families of weights @xmath19 where the methods of the present paper could be applied .
a difficulty here is that integral tqft as we need it in this paper is so far not developed for other tqfts . throughout the paper , we assume @xmath0 and we use the notation @xmath27 .
the construction of the modules @xmath3 goes as follows . for @xmath48 and @xmath49
, we construct certain simple modules which we denote by @xmath50 .
note that there are @xmath2 choices of pairs @xmath51 .
the construction of the modules @xmath50 is based on results from integral tqft obtained in @xcite . from the tqft description , we shall compute the dimension and weight space decomposition of @xmath50 . in particular , we shall compute the highest weight occuring in @xmath50 , thereby identifying @xmath50 with one of the @xmath13 in theorem [ 1.1 ] . , width=172 ] here is the construction of @xmath50 .
we give a description which can be read without any knowledge of tqft .
consider the graph @xmath52 depicted in figure [ lol ] which we call a lollipop tree .
it has @xmath53 trivalent vertices and one univalent vertex which in the figure is labelled @xmath54 .
the @xmath55-valent ` corner ' vertex to the left of the figure should be ignored , and the two edges meeting there are to be considered a single edge .
thus , @xmath52 has @xmath56 edges , @xmath8 of which are loop edges .
the edges incident to a loop edge are called stick edges , and we refer to a loop edge together with its stick edge as a lollipop .
a @xmath6-color is an integer @xmath57 .
a @xmath6-coloring of @xmath52 is an assignment of @xmath6-colors to the edges of @xmath52 .
a @xmath6-coloring is _ admissible _ if whenever @xmath58 , @xmath59 and @xmath60 are the colors of edges which meet at a vertex , then @xmath61 admissibility at the trivalent vertex of the @xmath58-th lollipop implies that the stick edge has to receive an even color , which we denote by @xmath62 , and the loop edge has to receive a color of the form @xmath63 , with @xmath64 .
we denote the colors of the remaining edges by @xmath65 as in figure [ lol2 ] , and we write an admissible @xmath6-coloring as @xmath66 .
, width=172 ] a @xmath6-coloring is of type @xmath51 if the color @xmath54 is assigned to the edge incident with the univalent vertex and if @xmath67 a @xmath6-coloring is _ small _ if the colors @xmath63 of the loop edges satisfy @xmath68 let @xmath69 denote the set of small admissible @xmath6-colorings of @xmath52 of type @xmath51 .
let @xmath70 denote the finite field with @xmath6 elements , and let @xmath71 be the @xmath70-vector space with basis @xmath69 .
there is an irreducible representation of the finite symplectic group @xmath72 on @xmath71 .
this will be proved in section [ sec3 ] .
steinberg s restriction theorem ( see _ e.g. _ @xcite ) implies that there is a unique simple @xmath4-module @xmath50 characterized by the following two properties : \(i ) the restriction of @xmath50 to the finite group @xmath72 is @xmath73 .
\(ii ) @xmath50 has @xmath6-restricted highest weight .
we recall that a dominant weight @xmath74 is _ @xmath6-restricted _ if , for each @xmath75 , we have @xmath76 . part ( i ) of the following theorem says that the @xmath50 are precisely the simple modules @xmath3 listed in theorem [ 1.1 ] .
part ( ii ) gives the weight space decomposition and thus determines the formal character of these modules .
to state the result , let @xmath77 be the multiset of weights occuring in @xmath50 .
( by a multiset , we mean a set with multiplicities . ) \(i ) the @xmath4-module @xmath50 is isomorphic to @xmath3 where the highest weight @xmath78 is given by @xmath79
\(ii ) we have @xmath80 where the weight of a coloring @xmath66 is @xmath81 in case @xmath82 , the highest weight corresponds to the @xmath6-coloring @xmath83 where all edges are colored zero .
indeed , formula ( [ defw ] ) gives @xmath84 in the other cases , the coloring @xmath83 is not allowed as it is not of type @xmath51 for @xmath85 .
we shall describe the colorings corresponding to the highest weights in case @xmath86 in section [ sec4 ] .
the @xmath45 case of theorem [ 1.8 ] answers affirmatively the question raised in ( * ? ? ? * p. 257
( after theorem 8.1 ) ) ( see also ( * ? ? ?
* ( after corollary 3 ) ) ) .
the remainder of this paper is organized as follows . in section [ sec2 ] ,
we formulate two results ( lemma [ elb1 ] and lemma [ elb2 ] ) about the @xmath72-modules @xmath71 . in section [ sec3 ] , we review the construction of @xmath71 and the proof of theorem [ 1.7 ] , and then prove lemma [ elb1 ] and lemma [ elb2 ] using further arguments from tqft . in section [ sec4 ] ,
we prove theorems [ 1.1 ] and [ 1.8 ] .
the only results from tqft that will be used in the proof of these two theorems are those stated in section [ sec2 ] .
finally , in section [ sec5 ] , we make a few further comments and discuss the rank 3 case as an example . * acknowledgements . * we thank henning h. andersen for helpful discussions .
he suggested checking our results against the jantzen sum formula in the rank @xmath87 case ( see section [ sec5 ] ) and showed us how to do it . g.
m. thanks the mathematics department of louisiana state university , baton rouge , the centre for quantum geometry of moduli spaces , aarhus , denmark , and the max planck institute for mathematics , bonn , germany , for hospitality while part of this paper was written .
p. g. also thanks the max planck institute for mathematics for hospitality .
last but not least , we thank the referee for his insightful comments .
we begin by fixing some notation . for @xmath60
any of the rings @xmath88 , @xmath70 , or @xmath5 , we take @xmath89 to be the subgroup of @xmath90 consisting of isometries of the skew symmetric form given by the matrix @xmath91 $ ] .
let @xmath92 be the maximal torus of @xmath4 given by the diagonal matrices of @xmath4 .
for @xmath93 and @xmath94 , let @xmath95 denote the diagonal matrix with @xmath96 on the @xmath97th diagonal entry , @xmath98 on the @xmath99-th diagonal entry , and @xmath100 s elsewhere on the diagonal .
we have an isomorphism @xmath101 we denote by @xmath102 the standard basis of the weight lattice @xmath103 where @xmath104 and @xmath105 for @xmath106 .
we also let @xmath107 denote the borel subgroup of @xmath4 which is the group of block matrices of the form @xmath108 where @xmath109 is an invertible upper triangular matrix and @xmath110 satisfies @xmath111 recall that @xmath71 is a representation of the finite symplectic group @xmath72 on the @xmath70-vector space with basis @xmath69 .
let @xmath112 denote the basis vector correponding to the coloring @xmath113 .
since the finite field @xmath70 is a subfield of @xmath5 , we may consider the actions of the finite maximal torus @xmath114 and of the finite borel subgroup @xmath115 on @xmath71 .
the following lemma says that the basis vectors @xmath112 are in some sense ` weight vectors ' for @xmath116 .
each basis vector @xmath117 is a simultaneous eigenvector for the commuting operators @xmath118 for @xmath119 , with eigenvalues given by @xmath120 where @xmath66 .
note that the exponents are meaningful only modulo @xmath2 , as @xmath121 is a cyclic group of order @xmath2 .
we may interpret the collection of these exponents for a given basis vector @xmath117 as specifying a reduced weight , by which we mean an element of @xmath122 let @xmath123 be the multiset of reduced weights occuring in @xmath71 .
the following is an immediate corollary of lemma [ elb1 ] .
we have @xmath124 where @xmath125 is the reduction modulo @xmath2 of @xmath126 as defined in ( [ defw ] ) . in section [ sec4 ]
, we shall see that this information is enough to determine @xmath77 , except that in case @xmath82 , we will also need to use the following lemma .
let @xmath127 be the coloring where all edges are colored zero .
then the basis vector @xmath128 in @xmath129 is fixed up to scalars by the finite borel subgroup @xmath130 .
the proofs of lemma [ elb1 ] and lemma [ elb2 ] will be given in section [ sec3 ] .
in section [ sec4 ] we shall apply lemma [ elb2 ] through the following corollary whose proof we give already here .
the highest weight of @xmath131 is congruent modulo @xmath2 to @xmath132 .
let @xmath133 be a highest weight vector in @xmath131 .
then @xmath133 is fixed up to scalars by @xmath107 ( see @xcite ) . restricting to the finite symplectic group
, we can view @xmath133 as a vector in @xmath73 that is fixed up to scalars by @xmath130 . by (
* theorem 7.1 ) , there is a unique line fixed by @xmath130 in @xmath73 . since @xmath128
is also contained in this line by lemma [ elb2 ] , we conclude that @xmath133 and @xmath128 are proportional .
in particular , @xmath133 and @xmath128 have the same reduced weight , which implies the result .
in this section , we review how integral tqft leads to the irreducible @xmath72-representations @xmath71 of theorem [ 1.7 ] which were the starting point for this paper .
in particular , we show how theorem [ 1.7 ] follows from @xcite using a result ( lemma [ lemm - app ] ) originally proved in @xcite .
we shall provide a self - contained proof of lemma [ lemm - app ] in appendix [ app - a ] .
we then prove lemma [ elb1 ] and lemma [ elb2 ] .
let @xmath134 denote a closed surface of genus @xmath8 equipped with one marked framed point labelled @xmath54 , where @xmath38 is an integer with @xmath135 .
( recall @xmath27 . )
integral tqft @xcite associates to @xmath134 a free @xmath136$]-module @xmath137 of finite rank , together with a projective - linear representation of the mapping class group of @xmath134 on this module . here
@xmath0 is a prime , @xmath138 is a primitive @xmath6-th root of unity , and @xmath136 $ ] is the ring of cyclotomic integers .
the mapping class group of @xmath134 can be identified with @xmath139 , that is , the mapping class group of @xmath140 , an oriented surface of genus @xmath8 with one boundary component .
( thus @xmath139 is the group of orientation - preserving diffeomorphisms of @xmath140 that fix the boundary pointwise , modulo isotopies of such diffeomorphisms . )
the projective - linear representation of @xmath139 on @xmath137 can be lifted to a linear representation of a certain central extension of @xmath139 .
the representations of mapping class groups obtained in this way may be considered as an integral refinement of the complex unitary representations coming from witten - reshetikhin - turaev tqft associated to the lie group @xmath21 . in particular , the rank of the free @xmath136$]-module @xmath137 is given by the verlinde formula ( [ verl ] ) .
recall that @xmath141 is a prime in @xmath136 $ ] , and @xmath136/(1-\zeta_p)$ ] is the finite field @xmath70 .
thus we get a representation on the @xmath70-vector space @xmath142 it is shown in ( * ? ? ?
12.4 ) that this induces a linear representation of @xmath139 on @xmath143 ( _ i.e. _ the central extension is no longer needed ) . furthermore , we proved in @xcite that @xmath143 has a composition series with ( at most ) two irreducible factors .
these irreducible factors are the @xmath71 defined in the introduction .
more precisely , we have a short sequence of @xmath139-representations @xmath144 it remains to show that the action of @xmath139 on the irreducible factors @xmath71 factors through an action of the finite symplectic group @xmath72 . for @xmath145 ,
this was proved by explicit computation in @xcite . for @xmath146 , we use the following lemma whose proof is deferred to appendix [ app - a ] .
the torelli group @xmath147 acts trivially on @xmath148 and @xmath149 .
it follows that the action of @xmath139 on the irreducible factors @xmath71 factors through an action of the symplectic group @xmath150 to see that this descends to an action of the finite symplectic group @xmath72 , we invoke a result of mennicke , who proved that for @xmath146 , the group @xmath72 is the quotient of @xmath151 by the normal subgroup generated by the @xmath6-th power of a certain transvection ( * ? ? ?
* satz 10 ) .
the result follows , because transvections lift to dehn twists in @xmath139 , and it is well - known that in @xmath21-tqft at the prime @xmath6 the @xmath6-th power of any dehn twist acts trivially .
this concludes the proof of theorem [ 1.7 ] . for the proof of lemma [ elb1 ] and lemma [ elb2 ] , we need to say more about the basis vectors @xmath117 associated to colorings @xmath152 .
recall the graph @xmath52 depicted in figure [ lol ] .
a regular neighborhood in @xmath153 of @xmath52 is a @xmath87-dimensional handlebody @xmath154 .
we identify @xmath134 with the boundary of @xmath154 , in such a way that the univalent vertex labelled @xmath54 in the figure meets the boundary surface in the marked point . given this identification ,
there is a basis @xmath155 of @xmath137 called the orthogonal lollipop basis ( see @xcite ) .
the basis vectors are indexed by colorings @xmath152 in @xmath156 .
reducing modulo @xmath141 , we get a basis @xmath157 of @xmath143 .
notice that as an @xmath70-vector space , @xmath143 is the direct sum of @xmath148 and @xmath149 . the basis vectors @xmath112 where @xmath113 are a basis of @xmath71 . before giving the proof of lemma [ elb1 ] and lemma [ elb2 ]
, we also need to fix our conventions for the homomorphism @xmath158 .
this homomorphism comes from the action of the mapping class group @xmath139 on the homology of the surface @xmath134 by isometries of the intersection form .
recall that we have identified @xmath134 with the boundary of a regular neighborhood @xmath154 of the graph @xmath52 .
we identify @xmath159 with @xmath160 identifying the homology class of the positive meridian of the @xmath58th loop ( oriented counterclockwise and counting from the left to right ) with the @xmath58th basis vector of @xmath160 and denote this element by @xmath161 .
similarly we identify the homology class of the parallel to the @xmath58th loop to be the @xmath162th basis vector of @xmath160 .
then the intersection pairing is described by the matrix @xmath91 $ ] .
first , let us prove lemma [ elb1 ] in the special case when @xmath145 .
then @xmath163 is zero and @xmath164 has dimension @xmath165 , as the graph @xmath166 is just a single lollipop , with stick color @xmath167 equal to @xmath54 , so that only the color @xmath168 of the loop edge may vary , and there are @xmath169 possibilities for @xmath170 .
the representation of @xmath171 on @xmath172 is shown in @xcite to be isomorphic to the standard representation of @xmath173 on homogeneous polynomials of degree @xmath174 in two variables , say @xmath175 and @xmath176 .
explicitly , this representation is given by : @xmath177 note that @xmath178 thus @xmath179 , which is the @xmath180-th element in the monomial basis for the polynomials , is an eigenvector for @xmath181 $ ] with eigenvalue @xmath182 .
one can check that the intertwiner @xmath183 @xcite defining the isomorphism sends @xmath179 to a multiple of @xmath117 for @xmath152 the coloring which is @xmath54 on the stick edge and @xmath184 on the loop edge of @xmath166 .
thus lemma [ elb1 ] holds when @xmath145 . the general case is now proved as follows .
the torus @xmath116 is contained in the subgroup of @xmath185 isomorphic to a product of @xmath8 copies of @xmath186 arising from the @xmath8 copies of @xmath187 in @xmath188 corresponding to each loop of @xmath52 . specifically , the element @xmath189 defined in section [ sec2 ] lies in the @xmath58-th copy .
similarly , the mapping class group @xmath139 contains a subgroup isomorphic to a product of @xmath8 copies of @xmath190 .
the @xmath58-th copy of @xmath190 is generated by the dehn twist about the meridian and the dehn twist about the parallel to the @xmath58th loop of @xmath52 .
since the homomorphism @xmath191 is surjective , we can lift @xmath189 ( non - uniquely ) to a mapping class , say @xmath192 , which we may assume to lie in the @xmath58-th copy of @xmath190 .
this copy of @xmath190 is the mapping class group of the one - holed torus which is cut off from @xmath134 by the simple closed curve @xmath193 on @xmath134 which is a meridian of the @xmath58-th stick edge of the graph @xmath52 . using the integral modular functor properties of ( * ? ? ?
* section 11 ) , we have an injective linear map @xmath194 given by gluing along @xmath193 . here
, @xmath195 stands for a genus @xmath196 surface with two marked points labelled @xmath62 and @xmath54 , respectively . the module
@xmath197 is again a free @xmath136$]-lattice by ( * ? ? ?
* theorem 4.1 ) .
the image of the gluing map ( [ if2 ] ) is a free sublattice of @xmath137 of full rank . when tensored with the quotient field of @xmath136 $ ] , the map ( [ if2 ] ) becomes an isomorphism familiar in tqfts defined over a field under the name of ` factorization along a separating curve ' . over
the ring @xmath136 $ ] the gluing map ( [ if2 ] ) is , however , not surjective in general . on the sublattice of @xmath137 given by the image of the map ( [ if2 ] )
, the mapping class @xmath192 preserves the direct sum decomposition and in each summand , @xmath192 acts only on the first tensor factor @xmath198 . when reduced modulo @xmath141 , the action induced by @xmath192 on @xmath199 is as described in the genus one case .
in particular , the lollipop with stick color @xmath62 and loop color @xmath200 indexes an eigenvector for the induced action of @xmath192 on @xmath199 with eigenvalue @xmath201 .
if the map ( [ if2 ] ) were an isomorphism of @xmath136$]-modules , this would prove the lemma by familiar tqft arguments , since it would then also induce an isomorphism when reduced modulo @xmath141 .
although ( [ if2 ] ) is not an isomorphism of @xmath136$]-modules , we are saved by the following fact ( see ( * ? ? ?
* theorem 11.1 ) ) . pick a basis @xmath202 of the lattice
@xmath197 associated to a lollipop tree as in ( * ? ? ?
* theorem 4.1 ) .
then the image under the map ( [ if2 ] ) of the direct summand @xmath203 of the l.h.s . of ( [ if2 ] ) is a certain power of @xmath141 times a direct summand of the r.h.s . of ( [ if2 ] ) , that is , of @xmath137 .
( the power of @xmath141 may depend on the summand . )
thus the action of @xmath192 on this direct summand , and hence the action of @xmath95 on the reduction modulo @xmath141 of this direct summand , can be computed from the action of @xmath192 on @xmath204 . since this action
is given by the genus one case , where the lemma is already proved , it follows that for a coloring @xmath66 , the basis vector @xmath205 is an eigenvector of @xmath95 with eigenvalue @xmath201 .
this completes the proof .
we now prepare the way for the proof of lemma [ elb2 ] .
let @xmath206 be the span of the homology classes of the meridians @xmath207 in @xmath208 .
note that @xmath206 is a lagrangian subspace with respect to the form @xmath209 .
let @xmath210 be the subgroup of @xmath188 consisting of the matrices which preserve @xmath206 .
one has that @xmath210 is the set of matrices of the form @xmath211 $ ] where @xmath212 , and @xmath110 satisfies @xmath213 we call this subgroup the lagrangian subgroup .
let @xmath214 be the mapping class group of the closed surface @xmath215 of genus @xmath8 , viewed as the boundary of the handlebody @xmath154 .
note that @xmath206 is the kernel of the map @xmath216 .
if @xmath217 , and @xmath218 extends to a diffeomorphism @xmath219 then @xmath220 .
we have a converse : if @xmath221 , then @xmath218 is induced by an element of @xmath214 which extends to a diffeomorphism @xmath219 .
consider the special case when @xmath221 has the form @xmath222 $ ] .
it follows that @xmath223 consider the building blocks @xmath224 $ ] where @xmath225 has zero entries everywhere except for the @xmath226 location where it has a @xmath100 , and @xmath227 ( for @xmath228 ) has zero entries everywhere except for the @xmath229 location and the @xmath230 location where it has @xmath100 s . these @xmath227 are realized by dehn twists along @xmath161 in the case @xmath231 , and along a curve representing @xmath232 when @xmath228 .
these curves may be chosen so that they bound disks in @xmath154 .
thus these dehn twists extend over @xmath154 .
products of such dehn twists realize any symmetric matrix @xmath110
. see ( * ? ? ?
* p 312 - 313 ) .
we can reduce the general case to the above case , using another special case : @xmath221 has the form @xmath233 $ ] .
we note that this is the case when @xmath234 where @xmath235 . an elementary matrix in @xmath236
can be realized , as @xmath237 , by sliding one 1-handle in @xmath154 over another .
permuting two handles realizes , as @xmath237 , a transposition matrix .
any @xmath235 is a product of elementary matrices and perhaps a transposition matrix .
thus @xmath238 $ ] for any @xmath235 can be realized by a diffeomorphism which extends over @xmath154 .
we let @xmath239 denote the unipotent radical of the finite borel subgroup @xmath130 . the image of @xmath210 under the quotient map @xmath240 contains @xmath239 .
we have that @xmath239 is the group of block matrices over @xmath70 of the form @xmath241 $ ] where @xmath242 is an invertible upper triangular matrix with @xmath100 s on the diagonal and @xmath110 satisfies @xmath243 each such matrix may be factored @xmath244 \left[\begin{smallmatrix } v&0 \\ 0 & ( v^t)^{-1}\end{smallmatrix } \right],$ ] and @xmath245 will equal its transpose . as above
, we note that @xmath244 $ ] can be written as a product of @xmath246 $ ] matrices .
thus any matrix of the form @xmath244 $ ] has lifts under the quotient map @xmath240 that lie in @xmath210 .
also any matrix of the form @xmath247 $ ] has such a lift .
it follows that any element of @xmath239 lifts to an element of @xmath210 .
[ proof of lemma [ elb2 ] ] recall that @xmath127 denotes the coloring where all edges are colored zero .
we are to show that the basis vector @xmath128 in @xmath129 is fixed up to scalars by @xmath130
. it will suffice to show that @xmath128 is fixed by @xmath239 .
note that @xmath128 is the reduction modulo @xmath141 of the basis vector @xmath248 of @xmath249 which is represented by the empty skein , and is thus fixed by any element of @xmath250 which extends to a diffeomorphism of @xmath154 , as observed in remark [ 3.3 ] . by proposition [ lzf ] , any element of @xmath239 lifts to an element of @xmath210 , and by proposition [ lextend ] , any element of @xmath210 is induced by an element of @xmath214 which extends to a diffeomorphism of @xmath154 .
this implies the result .
theorem [ 1.1 ] follows easily from theorem [ 1.8 ] and the dimension formulae of @xcite , where we computed the cardinality of the sets @xmath69 in the various cases in terms of the verlinde formula ( [ verl ] ) and its cousin ( [ d - verl ] ) . in the proof of theorem
[ 1.8 ] , we shall need one more result from modular representation theory . recall that the set of simple positive roots for the symplectic lie algebra consists of @xmath251 , and @xmath252 ( * ? ? ?
* planche iii ) .
let @xmath253 and suppose @xmath19 is a @xmath6-restricted dominant weight for @xmath4 .
let @xmath254 be the set ( without multiplicities ) of weights occuring in the simple module @xmath13 . 1 .
if @xmath255 is such that @xmath256 for all @xmath25 , then @xmath257 .
2 . if @xmath258 and @xmath259 for some @xmath25 , then @xmath260 .
this is true for the sets of weights of simple @xmath12-modules . by a result of premet @xcite
( see also the discussion in @xcite ) , @xmath254 is the same when working over @xmath5 or @xmath261 as long as @xmath19 is @xmath6-restricted and the characteristic @xmath253 .
the result follows .
let us now prove theorem [ 1.8 ] . recall that we must determine @xmath77 (= the multiset of weights occuring in @xmath50 ) , and we must determine which of the weights in @xmath77 is the highest weight , which we denote by @xmath262 .
as the details of this are somewhat involved , let us first outline the strategy of the proof .
the proof proceeds in four steps , as follows .
* the first step is to compute @xmath77 modulo @xmath2 . as observed in remark [
2.3 ] , we already know the answer : it is the multiset @xmath123 which was determined in corollary [ 2.2 ] . recall that the elements of @xmath123 are reduced weights , and that every coloring in @xmath69 determines a reduced weight in @xmath123 .
* step 2 . *
the second step is to identify @xmath263 , that is , the reduced weight in @xmath123 which is the reduction modulo @xmath2 of the highest weight . in case @xmath82 , this is the reduced weight associated to the zero coloring , as proved in corollary [ 2.4 ] . in the three other cases , we will show that @xmath263 is the reduced weight associated to a coloring illustrated in figure [ all ] .
, , case @xmath264 , and case @xmath40 .
the leftmost part of @xmath52 is not drawn as it is colored zero .
for the same reason the rightmost edge is not drawn in case @xmath40 as @xmath38 is zero .
these graphs could be guessed by taking the smallest coloring which is rightmost on the graph and has the given type.,title="fig:",width=115 ] , , case @xmath264 , and case @xmath40 .
the leftmost part of @xmath52 is not drawn as it is colored zero .
for the same reason the rightmost edge is not drawn in case @xmath40 as @xmath38 is zero .
these graphs could be guessed by taking the smallest coloring which is rightmost on the graph and has the given type.,title="fig:",width=115 ] this will be proved by showing that these colorings satisfy the hypothesis of the following lemma .
we let @xmath265 denote the reduction modulo @xmath2 of the root @xmath266 .
let @xmath152 be a coloring in @xmath69 and let @xmath125 be its associated reduced weight . if @xmath267 for all @xmath25 , then @xmath268 .
let @xmath269 be any lift of @xmath125 to @xmath77 .
the hypothesis implies that for all @xmath25 , the weight @xmath270 does not occur in @xmath77 . by the construction of the module @xmath50 , we know that its highest weight is @xmath6-restricted , so that we can apply lemma [ lem - premet](i ) .
it follows that @xmath269 is the highest weight , and so @xmath125 is the reduction modulo @xmath2 of the highest weight , as claimed
. * step 3 .
* once @xmath271 is known , the third step will be to determine @xmath272 .
write @xmath273 with @xmath274 ( since @xmath262 is @xmath6-restricted ) .
the coefficient @xmath275 is determined by its reduction @xmath276 modulo @xmath2 provided @xmath277 .
but if @xmath278 , then @xmath275 can be either @xmath279 or @xmath2 . we shall show that @xmath280 whenever @xmath278 in all our cases by means of the following lemma .
let @xmath281 be the highest weight in @xmath77 .
if @xmath282 , then @xmath280 .
this follows immediately from lemma [ lem - premet](ii ) .
* once step 3 is completed , it only remains to prove that @xmath77 is as claimed in the theorem .
this is now easy .
recall the notation @xmath27 .
we simply note that all weights in @xmath77 must lie in @xmath283^g$ ] as they must lie in the convex hull of the orbit under the weyl group of the highest weight , and in each case this highest weight has been shown in step 3 to lie in @xmath284^g$ ] .
but no two distinct integer points in @xmath283^g$ ] agree modulo @xmath285 in each coordinate .
thus @xmath77 is determined by its reduction modulo @xmath2 , and we are done . in the rest of this section , we shall now carry out steps 2 and 3 in the various cases .
having done this , the proof will be complete . to simplify notation
, we shall denote the highest weight @xmath262 simply by @xmath19 . also , from now on when we say coloring , we mean a small admissible @xmath6-coloring .
recall that a coloring @xmath66 assigns the color @xmath62 to the @xmath58th stick edge and the color @xmath63 to the @xmath58th loop edge ( see figure [ lol2 ] ) .
recall also that @xmath286 , @xmath287 satisfy the smallness condition @xmath288 , and the coefficient of @xmath289 in the weight @xmath290 is @xmath291 ( see ( [ defw ] ) ) . before we begin with the cases , we state two lemmas .
both are an easy consequence of corollary [ 2.2 ] and the smallness condition .
recall @xmath292 if @xmath293 for some @xmath294 , then @xmath295 , suppose @xmath66 , and @xmath296 .
if @xmath297 for some @xmath294 , then @xmath298 * case @xmath82 .
* recall @xmath299 is the coloring which is zero on every edge .
step 2 was already taken in corollary [ 2.4 ] and so we know that @xmath300 . in step 3 , we must show that @xmath301 . by lemma
[ 4.4 ] , it is enough to show that @xmath302 for @xmath303 , which follows easily from lemma [ dcoef ] .
this completes the proof in case @xmath82 .
* case @xmath304 . *
let @xmath152 be the coloring on the left of figure [ all ] , then @xmath305 we shall show that @xmath306 as explained above , step 2 in the proof is based on lemma [ 4.1 ] .
we must show that @xmath307 for all @xmath75 . for @xmath308
this follows from lemma [ dcoef ] . for @xmath309 ,
it is proved by contradiction , as follows .
assume that @xmath310 .
then there is a coloring @xmath311 of type @xmath312 with @xmath313 we will see such a coloring is impossible . by lemma [ d-1 ] , @xmath314 for @xmath315 . in other words
, @xmath316 must color all but the rightmost lollipop by zero . by admissibility
, it follows that @xmath317 . on the other hand , since the coefficient of @xmath318 in @xmath319 is @xmath320 , we have @xmath321 so @xmath322 so @xmath323 by smallness and hence @xmath324 again by smallness ( see ( [ smallness ] ) ) .
this contradicts @xmath325 .
thus @xmath316 does not exist and hence @xmath326 .
this completes step 2 , and we now know that @xmath327 . next , step 3 in the proof is based on lemma [ 4.4 ] . we must show that @xmath328 whenever the coefficient of @xmath329 in @xmath330 is zero . for @xmath331 ,
this follows from lemma [ dcoef ] .
then if @xmath332 , step 3 is already complete , as the coefficient of @xmath333 and of @xmath334 in @xmath330 is non - zero .
but if @xmath335 , the coefficient of @xmath333 in @xmath330 is zero and we therefore also need to show that @xmath336 .
this is again proved by contradiction .
assume that @xmath337 .
then there is a coloring @xmath338 of type @xmath339 with @xmath340 as before when we ruled out the coloring @xmath341 in step 2 , it follows from lemma [ d-1 ] that @xmath338 must color all but the rightmost lollipop by zero . by admissibility
, it follows that @xmath342 and so @xmath343 by smallness .
but then the coefficient of @xmath318 in @xmath344 must be @xmath345 , which contradicts ( [ eq ] ) where this coefficient is @xmath346 .
this contradiction shows that @xmath338 does not exist .
this completes the proof in case @xmath304 . in case @xmath264 and @xmath40
, we will also need the following lemma which says that odd colorings must assign nonzero colors to at least two lollipop sticks , and to at least three lollipop sticks if @xmath39 .
suppose @xmath66 is a small admissible @xmath6-coloring of type @xmath347 . 1 .
there are at least two distinct @xmath348 with @xmath349 .
2 . if moreover @xmath39 , then there are at least three distinct @xmath348 with @xmath349 .
\(i ) if only one of the @xmath350 is non - zero , say @xmath351 , then admissibility implies that @xmath352 and so condition ( [ type ] ) in the definition of colorings of type @xmath347 is violated .
\(ii ) if moreover @xmath39 , and only two of the @xmath350 are non - zero , say @xmath351 and @xmath353 , then admissibility implies that @xmath354 and so condition ( [ type ] ) is again violated .
* case @xmath264 .
* let @xmath152 be the coloring in the middle of figure [ all ] , then @xmath355 we shall show that @xmath356 in step 2 , we must show that @xmath357 for all @xmath75 . for @xmath358
this follows from lemma [ dcoef ] . for @xmath359
it is proved by contradiction , as follows .
( case @xmath360 . )
assume that @xmath361 .
then there is a coloring @xmath316 of type @xmath347 with @xmath362 note @xmath316 must color all but the rightmost lollipop with zeros by lemma [ d-1 ] , and so @xmath316 contradicts lemma [ 3-lolli](i ) .
( case @xmath309 . )
assume that @xmath363 .
then there is a coloring @xmath338 of type @xmath347 with @xmath364 by lemma [ d-1 ] we have @xmath365 for @xmath366 . by admissibility
, it follows that the three colors meeting at the trivalent vertex at the bottom of the rightmost stick are @xmath367 and @xmath54 .
computing the coefficient of @xmath368 , we have @xmath369 or @xmath370 . by smallness
, it follows that @xmath371 .
this implies two things .
first , since @xmath372 ( _ i.e. , _ the coloring is odd ) , it follows that @xmath373 .
second , by the triangle inequality in the admissibility condition at the trivalent vertex at the bottom of the rightmost stick , it follows that @xmath374 .
one concludes that @xmath375 .
the rest of the proof is now the same as in case @xmath304 step 2 .
computing the coefficient of @xmath376 exactly as was done there ( see ( [ eqg ] ) ) , we deduce @xmath377 , hence @xmath378 and so @xmath379 by smallness .
this contradicts @xmath380 .
thus step 2 is complete , and we now know that @xmath381 for step 3 , we must show that @xmath328 whenever the coefficient of @xmath329 in @xmath330 is zero . for @xmath382 ,
this follows from lemma [ dcoef ] . then if @xmath383 , step 3 is already complete , as the coefficient of @xmath333 , @xmath334 , and @xmath384 in @xmath330 is non - zero .
but if @xmath385 , then we also need to show that @xmath386 , and if @xmath335 , then we also need to show that @xmath387 .
the arguments in these two cases will be given below .
note that for @xmath388 ( which corresponds to @xmath45 ) we have both @xmath385 and @xmath335 , so that we need to use both arguments .
( case @xmath385 . )
assume for a contradiction that @xmath389 .
then there is a coloring @xmath390 of type @xmath391 with @xmath392 by lemma [ d-1 ] @xmath390 must color all but the @xmath393st lollipop by zero , which violates lemma [ 3-lolli](i ) .
this shows that @xmath386 .
( case @xmath335 . )
assume for a contradiction that @xmath394 .
then there is a coloring @xmath395 of type @xmath396 with @xmath397 by the exact same reasoning as when showing that @xmath375 for the coloring @xmath338 in step 2 ( case @xmath309 ) , we have @xmath398 . hence @xmath399 ( since we assume @xmath335 ) and so @xmath400 by smallness .
but then the coefficient of @xmath376 in @xmath401 must be @xmath402 which contradicts ( [ eqtilde ] ) .
this shows that @xmath387 .
the proof in case @xmath264 is now complete .
* case @xmath40 .
* let @xmath152 be the coloring on the right of figure [ all ] , then @xmath403 we shall show that @xmath404 in step 2 , we must show that @xmath307 for all @xmath75 . for @xmath405 and also for @xmath309 , this follows from lemma [ dcoef ] . for @xmath406
, it is proved by contradiction , as follows .
assume that @xmath407 .
then there is a coloring @xmath316 of type @xmath408 with @xmath409 note that @xmath316 must color all but two of the lollipops with zeros by lemma [ d-1 ] and so @xmath316 contradicts lemma [ 3-lolli](ii ) .
this shows that @xmath410 .
we refer to this argument as the `` two lollipop argument '' . for @xmath360 ,
the proof is similar : we have that @xmath411 and the two lollipop argument shows that @xmath412 .
thus step 2 is complete , and we now know that @xmath413 concerning step 3 , we have that @xmath414 for @xmath415 by lemma [ dcoef ] , and for @xmath406 or @xmath196 by the two lollipop argument .
thus step 3 is complete , except if @xmath388 ( hence @xmath45 ) in which case we also need to show that @xmath416 , which is easy and left to the reader .
this completes the proof in case @xmath40 .
we elaborate on remark [ 1.6 ] .
let @xmath417 denote the closure of the fundamental alcove .
( see _ e.g. _ @xcite . )
a dominant weight @xmath9 lies in @xmath417 iff @xmath418 where @xmath419 is sum of the fundamental weights , and @xmath420 is the highest short root ( thus @xmath421 is the highest root of the dual root system ) . by ( * ? ? ?
* planche ii ) , we have @xmath422 using @xmath423 , one can check that all the weights @xmath19 arising in theorem [ 1.1 ] lie outside of @xmath417 ( _ i.e. , _ one has @xmath424 ) except for the weight @xmath425 in rank @xmath20 ( for which @xmath426 ) .
in fact , as soon as the rank @xmath427 , one uniformly has @xmath428 for all the weights in our list .
thus the distance of our weights to the fundamental alcove grows linearly with the rank @xmath8 .
on the other hand , theorem [ 1.1 ] also holds in rank @xmath429 for those weights where it makes sense ( _ cf .
_ remark [ 17s](i ) ) .
it turns out that those weights @xmath19 in rank @xmath429 all lie in @xmath417 .
hence the dimension of @xmath13 is given by the weyl character formula , as it is a well - known consequence of the linkage principle that for dominant weights @xmath19 in @xmath417 , the simple module @xmath13 is isomorphic to the weyl module @xmath430 ( see _ e.g. _ @xcite ) .
we have checked that indeed for @xmath429 our dimension formulae ( see appendix [ app - b].2 ) agree with the weyl character formula .
a further consistency check is possible in rank @xmath20 . in this case , although our weights ( with one exception ) lie outside of @xmath417 , the distance to @xmath417 is not too big ( one has @xmath431 ) and one can use the jantzen sum formula to compute the formal character of @xmath3 .
( see ( * ? ? ?
* ii.8 ) and references therein .
see also the summary in @xcite . ) here is the answer in the case @xmath432 we have @xmath433 one finds that @xmath13 is equal to the weyl module @xmath430 if @xmath434 , but for @xmath435 there is a short exact sequence @xmath436 where @xmath437 in particular @xmath22 can be computed from the weyl character formula as @xmath438 and we have checked that our dimension formulae ( see appendix [ app - b].3 ) agree with this . in rank @xmath439 , we have not attempted to compute @xmath13 with the jantzen sum formula .
note that it is easy to see that @xmath13 can only very rarely be equal to the weyl module @xmath430 for a weight @xmath19 that arises in theorem [ 1.1 ] , because of the following observations ( the first two of which imply the third ) . * by ( * ? ? ?
* corollary 2.10 ) , @xmath440 , with @xmath441 , @xmath38 , and @xmath442 held fixed and viewed as a function of @xmath6 is polynomial of degree @xmath443 .
* by the weyl character formula @xcite , @xmath444 , with @xmath445 , @xmath38 , and @xmath442 held fixed and viewed as a function of @xmath6 is polynomial of degree @xmath446 .
* for each @xmath447 , @xmath38 , and @xmath442 , there is a integer @xmath448 such that for all @xmath449 , @xmath450 .
in this appendix , we assume some familiarity with integral tqft , in particular with the results of @xcite and @xcite . see remark [ 3.2new ] above for the correspondence between our present notations and those in @xcite and @xcite .
it is shown in ( * ? ? ?
2.5 ) that every mapping class @xmath451 is represented on @xmath143 by a matrix of the form @xmath452 with respect to the direct sum decomposition ( as vector spaces ) @xmath453 ( here , the top left @xmath454 stands for an element of @xmath455 , and similarly for the two other @xmath454s . ) lemma [ lemm - app ] is equivalent to the following any @xmath218 in the torelli group @xmath147 is represented by a matrix of the form @xmath456 to prove this , let us review how the coefficients of this matrix can be computed . throughout this appendix we put @xmath457 .
recall that @xmath136 / ( h ) = { { \mathbb{f}}}_p$ ] .
we use the basis @xmath458 of @xmath143 ; it is the reduction modulo @xmath459 of the orthogonal lollipop basis @xmath460 of @xmath137 constructed in @xcite .
the basis @xmath460 is orthogonal with respect to the hopf pairing @xmath461 defined in @xcite , and we can therefore compute matrix coefficients by pairing with the dual basis @xmath462 satisfying @xmath463 here @xmath464 lies in the dual lattice @xmath465 .
we will use that @xmath466 where @xmath467 is defined in ( * ? ? ?
3.4 ) to be a certain power of @xmath459 times @xmath468 , and @xmath469 means equality up to multiplication by a unit in @xmath136 $ ] .
( the power of @xmath459 depends on @xmath152 . )
the hopf pairing is a symmetric @xmath136$]-valued form on @xmath137 which depends on a choice of heegaard splitting of @xmath470 .
given @xmath471 , their hopf pairing is computed skein - theoretically as follows .
think of @xmath96 and @xmath472 as represented by skein elements in the handlebody @xmath154 .
pick a complementary handlebody @xmath473 such that @xmath474 then @xmath475 where the right hand side of ( [ hdlb ] ) is the kauffman bracket evaluation ( at a certain square root of @xmath138 ) of the skein element in @xmath470 obtained as the union of @xmath96 in @xmath154 and @xmath476 in @xmath473 where @xmath477 is a certain identification of @xmath154 with @xmath473 ( see @xcite for a more precise definition ) .
given now a mapping class @xmath218 , the @xmath478 matrix coefficient of @xmath218 acting on @xmath137 is computed skein - theoretically as the evaluation @xmath479 in @xmath480 where @xmath481 is a certain skein element in @xmath482 obtained from @xmath218 in the usual way : there is a banded link @xmath483 in @xmath484 so that surgery on @xmath483 gives the mapping cylinder of @xmath218 ; one then obtains @xmath481 by replacing each component of @xmath483 by a certain skein element @xmath485 ( see @xcite ) and placing the resulting skein element in @xmath484 into @xmath482 in the standard way . here
@xmath483 and @xmath481 are not uniquely determined by @xmath218 , but it is shown in @xcite that this procedure is well - defined , and gives the correct answer .
( more precisely , it gives the correct answer up to multiplication by a global projective factor which is a power of @xmath138 .
here , we can safely ignore this projective ambiguity as we are eventually interested in the matrix of @xmath218 modulo @xmath457 only . )
we are now ready to prove lemma [ lemm - app-2 ] .
the main idea is that if @xmath218 lies in the torelli group @xmath147 , then @xmath483 and hence @xmath481 are very special , because the mapping cylinder of @xmath218 can be obtained by @xmath486-surgery on @xmath487 @xcite .
the notion of @xmath486-surgery goes back to matveev @xcite ( who called it borromean surgery ) , and then goussarov @xcite and habiro @xcite ( who called it clasper surgery ) .
we refer the reader to 5 of the survey @xcite for a good introduction to @xmath486-surgery and also for more references to the original papers .
the result we need is stated in ( * ? ? ?
5.5 ) and can be formulated as follows .
there is a certain 6-component banded link @xmath176 in a genus 3 handlebody with the following property . for every @xmath488 , there exists an embedding of a finite disjoint union of , say , @xmath489 copies of the pair @xmath490 into @xmath487 , giving rise to a @xmath491-component banded link @xmath483 in @xmath487 such that the mapping cylinder of @xmath218 is obtained by surgery on this banded link @xmath483 .
cabling each component of @xmath176 by @xmath485 gives rise to a skein element @xmath492 in @xmath493 , and the skein element @xmath481 appearing in our computation of matrix coefficients ( see ( [ mco ] ) ) is obtained by placing @xmath489 copies of @xmath492 into @xmath494 .
we can view @xmath492 as an element of the free @xmath136$]-module @xmath495 . the orthogonal lollipop basis @xmath460 of @xmath495 is indexed by colorings of the form @xmath496 . as before , let @xmath83 denote the zero coloring .
the following lemma is the key to proving lemma [ lemm - app-2 ] , as it shows that all but two coefficients of @xmath492 in the @xmath460 basis are divisible by @xmath459 .
@xmath497 for some @xmath498 $ ] .
we have that @xmath499 is represented by the empty link and @xmath500 is @xmath501 times the elementary tripod ( see ( * ? ? ?
* fig . 2 on p. 824 ) ) .
thus @xmath502 where @xmath503 is the disjoint union of @xmath60 elementary tripods embedded in @xmath504 .
the contribution of @xmath505 to the @xmath478 matrix coefficient of @xmath218 is @xmath506 for @xmath507 this is @xmath508 , as @xmath509 is the empty link . for @xmath510 , a straightforward application of the lollipop lemma ( * ? ? ?
7.1 ) shows that if @xmath152 and @xmath341 have the same parity , then ( [ mco2 ] ) is divisible by @xmath459 ( since a tripod has @xmath87 elementary lollipops ) .
this proves lemma [ lemm - app-2 ] .
the 6-component banded link @xmath176 in @xmath493 can be described as follows ( see ( * ? ? ?
5 ) ) . we can number the components of @xmath176 as @xmath511 such that the following holds .
the components @xmath512 lie in a ball @xmath513 and form zero - framed borromean rings . for @xmath514 , the component
@xmath515 is a zero - framed unknot going once around the @xmath58th ` hole ' of @xmath493 ; moreover @xmath515 is linked exactly once with @xmath516 , and unlinked with the two other @xmath517s .
recall that @xmath492 is obtained by cabling each of these 6 components by @xmath485 . for the lemma , we need to compute the coefficients of the basis vectors @xmath518 in @xmath492 .
( in fact , except for the zero coloring , we only need these coefficients modulo @xmath459 . )
such a coefficient is given by the hopf pairing @xmath519 as explained above , this is computed as the kauffman bracket of a certain skein element in @xmath470 .
using now the handle slide property of @xmath485 ( which is at the basis of the skein - theoretic construction of tqft @xcite ) , we can compute this by performing surgery on @xmath176 , which gives back @xmath470 , but transforms the standard embedding of @xmath520 in @xmath470 into an embedding @xmath521 of @xmath520 in @xmath470 where the three handles of @xmath522 are linked as in the borromean rings .
thus the coefficient of @xmath518 in @xmath523 is the kauffman bracket evaluation @xmath524 for the zero coloring , the basis vector @xmath525 is represented by the empty link , and so ( [ cov ] ) evaluates to @xmath100 , as asserted .
the following lemma [ yp2 ] shows that ( [ cov ] ) is divisible by @xmath459 as soon as @xmath526 or @xmath527 .
thus it only remains to compute ( [ cov ] ) for the colorings with @xmath528 and all @xmath529 . assume all @xmath529 . for @xmath530 , there is nothing to prove , while if one of the @xmath350 is zero , then ( [ cov ] ) evaluates to zero ( as the two remaining handles are unlinked ) .
this proves the lemma . for @xmath531
, we define its floor @xmath532 to be the largest integer @xmath533 , and its roof @xmath534 to be the smallest integer @xmath535 .
we have that ( [ cov ] ) is divisible by @xmath536 where @xmath537 ( @xmath514 ) , and w. l. o. g. we may assume @xmath538
. put @xmath539 then @xmath540 is represented by a linear combination of skein elements in @xmath493 with coefficients in @xmath136 $ ] .
( this follows from ( [ dl ] ) and the definition of @xmath541 in @xcite . ) below , we say that such a skein element has no denominators .
let @xmath420 denote the result of evaluating ( [ cov ] ) with @xmath540 in place of @xmath542 .
since the three lollipops are linked as in the borromean rings , we can pull the first two of them apart and consider the third as just some skein element in the complement of the first two .
in other words , by cutting @xmath540 at the midpoint of the edge labelled @xmath543 , we can write @xmath420 as the genus two hopf pairing of two skein elements , say @xmath544 and @xmath545 , in a genus @xmath55 handlebody with one banded point labelled @xmath543 . moreover , one of these two skein elements , say @xmath544 ( namely the one containing the first two lollipops ) , is a power of @xmath459 times the basis element @xmath546 of the orthogonal lollipop basis of @xmath547 . rewriting @xmath545 also in this basis and using the formulae in @xcite for the hopf pairing in the orthogonal lollipop basis
, one finds that @xmath420 is divisible by @xmath548 .
( here it is important to observe that both @xmath544 and @xmath545 are skein elements without denominators . )
thus ( [ cov ] ) is divisible by @xmath549 this completes the proof .
other applications of lemma [ yp2 ] include a skein - theoretic construction of ohtsuki s power series invariant for integral homology 3-spheres , and a torelli group representation inducing this invariant @xcite .
in ( * ? ? ? * prop . 7.7 , prop . 7.8 ) we gave residue formulae for @xmath42 and @xmath550 valid for @xmath445 . using equation ( [ diff ] ) , one
can then express @xmath551 , for the @xmath19 that arise in theorem [ 1.1 ] , for specified @xmath8 as polynomials in @xmath6 and @xmath38 using mathematical software .
below we write down these polynomials in rank @xmath429 , @xmath87 , and @xmath552 .
these formulae hold for @xmath0 and @xmath553 where @xmath27 .
the first polynomial in each rank is the second polynomial with @xmath38 set to zero .
similarly the fourth polynomial in each rank ( except rank 2 ) is the third polynomial with @xmath38 set to zero . as noted in section [ sec5 ] , our formulae in rank @xmath429 and also the first one in rank @xmath20 agree with weyl s character formula .
the second formula in rank @xmath20 agrees with weyl s character formula for @xmath385 , but not for @xmath554
. variations of knotted graphs .
the geometric technique of n- equivalence . in russian : algebra i analiz 12 ( 2000 ) , no . 4 , 79 - 125 .
english translation : st .
petersburg math .
j. 12 ( 2001 ) , no . 4 , 569 - 604 . from mapping class groups to monoids of homology cobordisms
: a survey . handbook of teichmller theory .
volume iii , 465 - 529 , irma lect .
phys . , 17 , eur .
soc . , zrich , 2012 .
arxiv.1003.2512 generalized surgeries of three - dimensional manifolds and representations of homology spheres . in russian : mat .
zametki 42 ( 1987 ) , no . 2 , 268 - 278 , 345 .
english translation : math
. notes 42 ( 1987 ) , no . 1 - 2 , 651 - 656 . | for @xmath0 a prime , and @xmath1 an integer , we use topological quantum field theory ( tqft ) to study a family of @xmath2 highest weight modules @xmath3 for the symplectic group @xmath4 where @xmath5 is an algebraically closed field of characteristic @xmath6 .
this permits explicit formulae for the dimension and the formal character of @xmath3 for these highest weights . |
the littlewood - richardson numbers @xmath2 for partitions @xmath3 are important in many areas of mathematics . for example , they are the structure constants of several related rings with distinguished bases : the ring of symmetric functions with its basis of schur functions , the representation ring of @xmath4 with its basis of irreducible highest weight modules , the external representation ring of the tower of symmetric groups with its basis of irreducible modules , and the cohomology ring of the grassmannian with its basis of schubert classes @xcite .
the combinatorics of littlewood - richardson numbers are extremely interesting and now we have many formulas for them , including the original littlewood - richardson formula @xcite . despite this deep and prolonged interest in littlewood - richardson numbers ,
one of the most fundamental questions about them was not asked until about a decade ago : @xmath5 this question came from ( of all places ) a problem in linear algebra concerning the possible eigenvalues of a sum of hermitian matrices .
the answer to this problem is given by the horn inequalities : the eigenvalues which can and do occur are the solutions to a set of linear inequalities , and the inequalities themselves come from non - negative integral eigenvalues solving this problem for smaller matrices .
the same inequalities answer our question about littlewood - richardson numbers .
a littlewood - richardson number @xmath2 is non - zero if and only if the triple of partitions @xmath6 satisfy certain linear inequalities , and the inequalities themselves come from triples indexing smaller non - zero littlewood - richardson coefficients .
this description is a consequence of work of klyachko @xcite which linked eigenvalues of sums of hermitian matrices , highest weight modules of @xmath4 , and the schubert calculus for grassmannians , and then knutson and tao s proof @xcite of zelevinsky s saturation conjecture @xcite .
this work implies horn s conjecture @xcite about the eigenvalues of sums of hermitian matrices .
these results have wide implications in mathematics ( see the surveys @xcite ) and have raised many new and evocative questions .
for example , the horn inequalities give the answer to questions in several different realms of mathematics ( representation theory , combinatorics , schubert calculus , eigenvalues ) , but it was initially mysterious why any one of these questions should have a recursive answer , as the proofs travelled through so many other parts of mathematics .
another question , which was the _ point de dpart _ for the results we describe here , is the following : are there related numbers whose non - vanishing has a similar recursive answer ?
our main result is a recursive answer for the non - vanishing of the analogs of littlewood - richardson coefficients for schur @xmath0-functions , and the same for schur @xmath1-functions .
we give one set of inequalities which determine non - vanishing for the @xmath0-functions and a different set of inequalities for the @xmath1-functions . because each schur @xmath0-function is a non - zero multiple of a corresponding schur @xmath1-function , a littlewood - richardson number for @xmath0-functions is non - zero if and only if the corresponding number for @xmath1-functions is non - zero , and thus our two sets of inequalities have the same sets of solutions .
another consequence of our work is a new set of recursive horn - type inequalities for the ordinary littlewood - richardson numbers @xmath2 . while these new inequalities are clearly related to the ordinary horn inequalities , they are definitely quite different .
( we explain this below . ) before we define some of these objects and give the different recursions , we remark that our results were proved using a mixture of the combinatorics of root systems , combinatorial linear algebra in lie algebras , and the geometry of certain flag varieties @xmath7 .
cominuscule flag varieties are ( almost all of ) the flag varieties whose geometrically defined bruhat order ( which is the bruhat order on the cosets @xmath8 of the weyl groups ) is a distributive lattice .
the alert reader will notice that these inequalities for schur @xmath0- and @xmath1-functions are not strictly recursive because they are indexed by ordinary littlewood - richardson numbers which are non - zero .
the reason for the term recursive is that the inequalities stem from a geometric recursion among all cominuscule flag varieties which is not evident from viewing only the subclass corresponding to , say the schur @xmath1-functions .
this abstract does not dwell on the geometry , but rather on the fascinating combinatorial consequences of these recursions .
the last section of this extended abstract does however give a broad view of some of the key geometric ideas which underly our recursion .
the results here are proved in the forthcoming preprint by the authors , `` the recursive nature of the cominuscule schubert calculus '' .
for more details and definitions concerning the various flavors of schur functions that arise here , we recommend the book of macdonald @xcite .
are symmetric functions indexed by , which are weakly decreasing sequences of nonnegative integers , @xmath9 .
the schur function @xmath10 is homogeneous of degree @xmath11 .
schur functions form a basis for the @xmath12-algebra of symmetric functions .
thus there are integral numbers @xmath2 defined by the identity @xmath13 homogeneity gives the necessary relation @xmath14 for @xmath15 .
a partition @xmath16 is represented by its , which is a left - justified array of boxes in the positive quadrant with @xmath17 boxes in row @xmath18 .
thus @xmath19 partitions are partially ordered by the inclusion of their diagrams .
let @xmath20 be the rectangular partition with @xmath21 parts , each of size @xmath22
. for @xmath23 ( @xmath24 and @xmath25 ) , define @xmath26 to be the partition obtained from the set - theoretic difference @xmath27 of diagrams ( placing @xmath16 in the upper right corner of @xmath20 ) .
thus we have @xmath28 the horn - type inequalities we give are naturally stated in terms of . for @xmath29 , define @xmath30 we say that a triple of partitions @xmath29 is if @xmath31 .
this convenient terminology comes from geometry .
suppose that @xmath23 and @xmath32 , where @xmath33 .
define @xmath34 draw @xmath16 in the upper right corner of the @xmath20 rectangle , and number the rows cartesian - style .
define @xmath35 to be the number of boxes that remain in @xmath16 after crossing out the rows indexed by @xmath36 .
[ ex : classichorn ] suppose that @xmath37 , @xmath38 , and @xmath39 , and we have @xmath40 and @xmath41 .
then the set @xmath42 is @xmath43 if we place @xmath16 in the upper - right corner of the rectangle @xmath44 and cross out the rows indexed by @xmath42 , @xmath45 we count the dots to see that @xmath46 .
a triple @xmath29 is feasible if and only if @xmath47 , and @xmath48 for all feasible triples @xmath49 and for all @xmath50 .
the first condition , @xmath47 , is due to homogeneity .
since replacing a partition @xmath16 by its conjugate @xmath51 ( interchanging rows with columns ) induces an involution on the algebra of symmetric functions , there is a version of the horn recursion where one crosses out columns instead of rows .
it turns out that there are necessary inequalities obtained by crossing out both rows and columns , including possibly a different number of each .
the cominuscule recursion reveals a sufficient subset of these .
[ definition : newhorn ] let @xmath52 and suppose that @xmath23 , @xmath32 , and we have another partition @xmath53 . define @xmath36 as before , and set @xmath54 draw @xmath16 in the upper right corner of the @xmath20 rectangle and cross out the rows indexed by @xmath36 and the columns indexed by @xmath55 . define @xmath56 to be the number of boxes that remain in @xmath16 .
we use the same data as in example [ ex : classichorn ] , and set @xmath57 . then @xmath58 if we now cross out the rows indexed by @xmath42 and the columns indexed by @xmath59 , @xmath60 we count the dots to see that @xmath61 .
a triple @xmath29 is feasible if and only if @xmath47 , and @xmath62 for all pairs of feasible triples @xmath49 and @xmath63 and for all @xmath64 .
the algebra of symmetric functions has a natural odd subalgebra which comes from its structure as a combinatorial hopf algebra @xcite .
this algebra was first studied by schur in the context of the theory of projective representations of the symmetric group .
this odd subalgebra has a pair of distinguished bases , the and the .
they are indexed by strict partitions , which are strictly decreasing sequences of positive integers @xmath65 .
they are proportional : @xmath66 , where @xmath16 has @xmath67 parts .
we have littlewood - richardson coefficients @xmath68 and @xmath69 indexed by triples of strict partitions and defined by the identities @xmath70 combinatorial formulas for these numbers were given in work of worley @xcite , sagan @xcite , and stembridge @xcite .
let @xmath71 be the strict partition of staircase shape .
then @xmath72 if @xmath73 .
if @xmath72 , define @xmath26 to be the partition obtained from the set - theoretic difference @xmath74 of diagrams ( placing @xmath16 in the upper right corner of @xmath75 ) .
thus we have @xmath76 as before , the horn - type inequalities are naturally stated in terms of symmetric littlewood - richardson numbers . for @xmath77 , define @xmath78 a triple of strict partitions @xmath29 is if @xmath79 .
we similarly define symmetric littlewood - richardson numbers @xmath80 for the schur @xmath0-functions . since the two bases are proportional , the corresponding coefficients are as well .
in particular the sets of triples @xmath3 for which the corresponding coefficients are feasible are the same .
nevertheless , we give two very different sets of inequalities which determine the feasibility of these numbers , arising from the different geometric origins of schur @xmath1-functions and schur @xmath0-functions .
[ def : typec ] let @xmath33 and suppose that @xmath81 is a strict partition and @xmath82 is an ordinary partition .
draw @xmath16 in the upper right corner of the staircase @xmath75 .
number the inner corners @xmath83 and , for each number in @xmath36 , cross out the row and column emanating from that inner corner .
then let @xmath84_\alpha}$ ] be the number of boxes that remain in @xmath16 .
[ def : typed ] let @xmath85 and suppose that @xmath81 is a strict partition and @xmath86 is an ordinary partition .
draw @xmath16 in the upper right corner of the staircase @xmath75 .
number the outer corners @xmath87 and for each number in @xmath88 , cross out the row and column emanating from the corresponding outer corner .
then let @xmath89 be the number of boxes that remain in @xmath16 .
[ ex : cd - horn ] for example , suppose that @xmath90 and @xmath91 , we have @xmath92 and @xmath93
. then @xmath94 and if we place @xmath16 in the upper - right corner of the rectangle and cross out the rows and columns emanating from the inner corners indexed by @xmath95 , we see that @xmath96_\alpha=6 $ ] .
if we instead cross out the rows and columns emanating from the outer corners indexed by @xmath97 , we see that @xmath98 .
the two diagrams are shown in figure [ fig : cd ] , on the left and right , respectively .
@xmath99_\alpha=6 $ } \put ( 4,10){\includegraphics{c.eps } } \end{picture } \qquad\qquad\qquad \begin{picture}(120,120)(0,15 ) \put(-4 , 20){1 } \put ( 9,33){\blue{2 } } \put ( 22 , 46){3 } \put(35 , 59){4 } \put ( 48,72){\blue{5 } } \put(61,85){\blue{6 } } \put(74 , 98){7 } \put(87,111){8 } \put(100,124){\blue{9 } } \put(-4,105){$\{\lambda\}_\alpha=5 $ } \put(12,10){\includegraphics{d.eps } } \end{picture}\ ] ] note that the homogeneity of the multiplication of schur @xmath0-functions and schur @xmath1-functions implies that @xmath100 is necessary for a triple @xmath101 to be feasible . a triple @xmath101 of strict partitions is feasible if and only if one of the following two equivalent conditions hold : \(1 ) the homogeneity condition holds , and for all feasible @xmath49 and all @xmath102 , we have @xmath103_\alpha + [ \mu]_\beta + [ \nu]_\gamma\ \leq\ \binom{n+1-r}{2}\,,\ ] ] or else
\(2 ) the homogeneity condition holds , and for all feasible @xmath104 and all @xmath105 with @xmath106 even , we have @xmath107
we first give some general idea of the ingredients in our proof , and then explain a little bit of the relation of this geometry to the combinatorics given here .
a flag manifold @xmath7 ( @xmath108 is a reductive algebraic group and @xmath0 is a parabolic subgroup ) has a bruhat decomposition into schubert cells indexed by cosets @xmath8 , where @xmath109 is the weyl group of @xmath108 and @xmath110 that of @xmath0 .
the closures of the schubert cells are the schubert varieties , and cohomology classes associated to them ( schubert classes ) form bases for the cohomology ring of @xmath7 .
standard results in geometry show that the structure constants ( generalized littlewood - richardson numbers ) are the number of points in triple intersections of general translates of schubert varieties ( and hence are non - negative ) .
if a structure constant is non - zero , then any triple intersection of corresponding schubert varieties ( not just a general intersection ) is non - empty , and general intersections are transverse .
conversely , if a structure constant is zero , then any corresponding general intersection is empty , and a non - empty intersection is never transverse .
the key idea is to replace the difficult question on non - emptiness of a general intersection of schubert varieties by the easier question of the transversality of a ( not completely general ) intersection .
this was used by one of us to show transversality of intersections in the grassmannian of lines @xcite , but its use to study the horn problem is due to belkale @xcite , who first gave a geometric proof of the horn recursion for the grassmannian .
this idea transfers the analysis from the flag manifold @xmath7 to its tangent space @xmath111 at a given point @xmath112 .
in fact , all of our diagrams are just pictures of the root - space decompositions of @xmath111 for the corresponding varieties . in our proof , we consider three schubert varieties which contain the point @xmath112 , and then move them independently by the stabilizer @xmath0 of @xmath112 so that they are otherwise general .
if it is possible to move the three tangent spaces inside of @xmath111 so that they meet transversally , then the triple is feasible , and if not , then it is not .
this explains where cominuscule flag manifolds come in .
the maximal reductive , or levi , subgroup @xmath113 of the parabolic group @xmath0 acts on the tangent space @xmath111 to @xmath7 at that point .
our arguments ( moving the tangent spaces to schubert varieties around by elements of @xmath113 ) require that @xmath113 act on @xmath111 with finitely many orbits , and this is one characterization of cominuscule flag manifolds @xmath7
. it also explains why there is a recursion .
the argument requires us to consider the stabilizer @xmath1 in @xmath113 of a certain linear subspace of @xmath111the tangent space to an orbit of @xmath113 through a general point of the intersection of general translates of the tangents to the three schubert varieties
. then the schubert calculus inside of the smaller flag manifold @xmath114 is used to analyze the transversality of that triple intersection .
fortunately , the flag manifold @xmath114 is itself cominuscule , which is the source of our recursion .
let @xmath115 be the grassmannian of @xmath21-planes in @xmath116 space .
the general linear group @xmath117 acts on @xmath115 . if @xmath118 then @xmath119 may be identified with the set of @xmath21 by @xmath22 matrices ( more precisely with @xmath120 ) .
the levi subgroup is the group @xmath121 which acts linearly on the rows and columns of @xmath21 by @xmath22 matrices .
the orbits of this group are simply matrices of a fixed rank , @xmath106 , and the subgroup @xmath1 is the stabilizer of a pair @xmath122 , where @xmath123 and @xmath124 both have dimension @xmath106 .
this explains why in definition [ definition : newhorn ] , the number of rows crossed out equals the number of columns crossed out .
the smaller cominuscule flag variety @xmath114 is the product of two grassmannians , @xmath125 .
fix a non - degenerate alternating bilinear ( symplectic ) form on @xmath126 .
let @xmath127 be the set of maximal isotropic ( lagrangian ) subspaces in @xmath126 , each of which has dimension @xmath21 .
this is the quotient of the symplectic group by the parabolic subgroup corresponding to the long root , @xmath128 .
since @xmath129 is isotropic the symplectic form identifies @xmath130 with the dual of @xmath131 , and @xmath132 is identified with the space of quadratic forms on @xmath131 .
the levi subgroup is the general linear group @xmath133 .
identifying @xmath131 with @xmath134 , the levi becomes @xmath135 and @xmath132 is the set of @xmath136 symmetric matrices .
( symmetric matrices are parametrized by their weakly lower triangular parts , which correspond to the staircase shape @xmath75 where the order of the columns has been reversed . )
the general linear group acts simultaneously on the rows and columns of symmetric matrices .
the orbits are simply symmetric matrices of a fixed rank , @xmath106 , and the subgroup @xmath1 is the stabilizer of the null space of such a matrix .
the smaller cominuscule flag variety @xmath114 is the grassmannian @xmath137 .
the schubert varieties of @xmath127 are indexed by strict partitions @xmath16 which fit inside the staircase @xmath75 , and its cohomology ring is the algebra of schur @xmath1-functions with these restricted indices . fix a non - degenerate symmetric bilinear form on @xmath138 .
the set of maximal isotropic subspaces ( each of which has dimension @xmath139 ) of @xmath138 has two isomorphic components .
let @xmath140 be one of these components .
this is the quotient of the even orthogonal group by a parabolic subgroup @xmath0 corresponding to one of the roots at the fork in the dynkin diagram , @xmath141 .
if @xmath142 is isotropic , then @xmath143 is identified with the space of alternating forms on @xmath131 .
the levi subgroup is the general linear group @xmath133 . identifying @xmath131 with @xmath144 ,
then the levi becomes @xmath145 and @xmath143 is the set of @xmath146 anti - symmetric matrices .
( anti - symmetric matrices are parametrized by their lower triangular parts , and these strictly lower triangular matrices correspond to the staircase shape where the order of the columns has been reversed . )
the general linear group acts simultaneously on the rows and columns of anti - symmetric matrices .
the orbits are simply anti - symmetric matrices of a fixed rank .
however , and this comes from the details of the proof and the roots of @xmath147 , the subgroup @xmath1 is the stabilizer of an even - dimensional subspace of @xmath131 .
the smaller cominuscule flag variety @xmath114 is the grassmannian @xmath148 , where @xmath106 is even .
the schubert varieties of @xmath149 are indexed by strict partitions @xmath16 which fit inside the staircase @xmath75 , and its cohomology ring is the algebra of schur @xmath0-functions with these restricted indices .
andrei zelevinsky , _ littlewood - richardson semigroups _ ,
new perspectives in algebraic combinatorics ( berkeley , ca , 199697 ) , math .
38 , cambridge univ . press , cambridge , 1999 , pp . | a consequence of work of klyachko and of knutson - tao is the horn recursion to determine when a littlewood - richardson coefficient is non - zero .
briefly , a littlewood - richardson coefficient is non - zero if and only if it satisfies a collection of horn inequalities which are indexed by smaller non - zero littlewood - richardson coefficients .
there are similar littlewood - richardson numbers for schur @xmath0- and @xmath1- functions . using a mixture of combinatorics of root systems , combinatorial linear algebra in lie algebras , and the geometry of certain cominuscule flag varieties , we give horn recursions to determine when these other littlewood - richardson numbers are non - zero .
our inequalities come from the usual littlewood - richardson numbers , and while we give two very different horn recursions , they have the same sets of solutions .
another combinatorial by - product of this work is a new horn - type recursion for the usual littlewood - richardson coefficients . |
there are a number of different astrophysical environments where optically thick material may be irradiated by x - rays .
if the incident spectrum has significant flux above 7.1 kev , then , as a result of its relatively high cosmic abundance and large fluorescent yield , the iron k@xmath0 line is predicted to be a significant feature in the resulting emission from the illuminated surface ( e.g. , * ? ? ?
* ; * ? ? ?
the presence of this line has been discussed for a number of different reprocessors including the solar surface @xcite , the surface of the companion star in x - ray binaries @xcite , and the surface of accreting magnetic white dwarfs ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
most famously , fe k@xmath0 emission has been observed in the x - ray spectra of many active galactic nuclei ( agn ) and black hole candidates @xcite . here , the irradiated surface is likely the accretion disc feeding the black hole , and some of the observed fe k@xmath0 lines have been observed to be broadened in a manner consistent with material orbiting in a relativistic gravitational potential @xcite .
finally , tentative detections of line emission , most likely from fe k@xmath0 , have been found in recent observations of the x - ray afterglows of some @xmath4-ray bursts ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
if this is confirmed then it implies that optically thick material may be in the vicinity a possible constraint on the environment and progenitor of the burst @xcite . as one of the few physical diagnostics in the x - ray spectra of these systems , it is important to understand how the fe k@xmath0 emission line depends on the properties of the irradiated gas .
previous studies have examined the influence of changes in metal abundance @xcite , the inclination angle of the reflector @xcite , ionization state @xcite , and the density distribution of the gas @xcite on the line emission .
this paper generalizes the results of matt and colleagues by presenting how the strength of the fe k@xmath0 line ( presented as an equivalent width [ ew ] ) depends on both the fe abundance and the ionization state of the gas .
these two variables have the greatest impact on the ew , and are likely to vary widely over different astrophysical environments .
we anticipate that these results will be helpful in interpreting fe k@xmath0 detections whenever it arises from an optically thick surface .
the following section details the calculations that were performed , and then section [ sect : res ] presents the results as contour plots of ew .
a brief discussion concludes the paper in section [ sect : discuss ] .
the calculations were performed using the code of @xcite .
a slab with a constant hydrogen number density @xmath510@xmath6 @xmath7 was illuminated with a power - law continuum with photon - index @xmath1 ( i.e. , so that photon flux @xmath8 ) .
provided that they have comparable flux above 7 kev , other incident spectral forms ( e.g. , blackbody , wien tail ) will not significantly alter the results . as a crude approximation to isotropic illumination , the penetrating radiation was assumed to strike the slab at an angle of 54.7 degrees from the normal .
hydrogen and helium were assumed to be fully ionized , but c v vii , o v ix , mg ix xiii , si xi xv , and fe xvi
xxvii were treated with abundances given by @xcite .
the depth of the slab was increased as the illuminating flux was increased , and had a minimum value of 6 thomson depths and a maximum of 30 thomson depths . when the irradiated material reached thermal and ionization equilibrium , the angle - averaged reprocessed emission from the gas was calculated .
as with all photoionized gas , the reflection spectrum can be characterized by an ionization parameter : @xmath9 where @xmath10 is the incident flux between 0.01 and 100 kev , the energy range of the illuminating continuum .
@xmath11 was varied by changing only the value of @xmath10 .
the evolution of the reflection spectrum with @xmath11 has been discussed by @xcite . here
we calculate reflection spectra for @xmath126.0 while also varying the fe abundance from 0.110.1 times solar .
when the illuminated gas is predominantly neutral , the fe k@xmath0line may be affected by variations in other metals , particularly oxygen @xcite . of course , when @xmath13 all the metals other than fe are fully ionized and will have little effect on the fe emission lines .
therefore , we do not consider abundance effects due to other metals , and instead refer readers to @xcite where such results are presented for neutral reflectors .
the iron emission features will also depend slightly on the slope of the incident continuum ( due to the different number of ionizing photons ) , so we have performed these computations for @xmath14 , 2.0 and 2.3 to illustrate this effect .
the ews were calculated by the following integral : @xmath15 where @xmath16 is the spectral flux of the reflected spectrum , and @xmath17 is the estimated spectral flux of the continuum at energy @xmath18 .
the continuum was estimated by fitting a straight line to the reflection spectrum between 5.50 and 7.11 kev .
the lower limit of the integration was chosen to take into account any compton broadening of the line profile , and the upper limit was set to the energy of the photoelectric absorption edge of neutral fe . dividing the line - flux by the monochromatic continuum at 6.4 or 6.7 kev makes a negligible difference to the contour plots .
figure [ fig : ew - ill - nor ] shows several examples of this calculation at different @xmath11 .
continuum was incident on the slab .
the solid lines show the reprocessed emission and the dotted lines are the estimate to the local continuum about the fe k@xmath0 line .
the ew of the line is then calculated by integrating the flux above the dotted line between 5.5 and 7.1 kev.,scaledwidth=50.0% ]
figure [ fig : contourplots - nor ] shows contours of constant fe k@xmath0 ew in the @xmath19-fe abundance plane for the case where the incident power - law has not been added to the reflected spectra , as if the reflected spectrum were viewed in isolation .
therefore , these ews are the _ maximum _ that can be observed as they are not diluted by the incident power - law . such reflection - dominated spectra may be observed in seyfert 2 galaxies ( e.g. , ngc 6552 ; @xcite ) .
the plots show that the fe k@xmath0 ew reaches two separate maxima in the vertical direction : one when the ionization parameter is low and the line is at 6.4 kev , and the other at @xmath20 when the line is at 6.7 kev . between these two maxima the line
is suppressed due to auger destruction @xcite , and at higher @xmath11 the line weakens and disappears as all the iron becomes ionized .
increasing the iron abundance does increase the fe k@xmath0 ew , but , as is discussed in sect .
[ sect : discuss ] , its effect is greatest when the abundance is still relatively low . nevertheless , fig .
[ fig : contourplots - nor ] shows that the line can get very strong ( @xmath21 kev in some cases ) at high fe abundance with the largest values occurring when the line is neutral . as is shown below , this is reversed when the incident power - law is added to the reflection spectrum . as @xmath1
is increased the number of fe ionizing photons decreases .
this causes the contours to spread out vertically in the plot , because at a higher @xmath1 a larger flux is needed to ionize fe .
when @xmath22 and @xmath23 the dependence on the fe abundance weakens and the ew grows more slowly .
again , this is due to the decrease in the number of ionizing photons . in many x - ray observations unprocessed radiation with the spectrum of the incident power - law
is simultaneously detected along with the reflection spectrum .
therefore , we have repeated all the ew calculations for the total power - law+reflection spectra ( agn workers think of this as a reflection fraction of unity ) .
these results are shown in figure [ fig : contourplots ] .
as before , two maxima are found in the contours : one when iron is neutral and the other when the he - like fe k@xmath0 emission line at 6.7 kevis prominent . likewise , the contours spread out vertically as the photon - index is increased
. however , there are a number of differences between this set of plots and the previous one .
the first one is trivial but important : the fe k@xmath0 ews are now much smaller .
the largest values are now just over 800 ev and occur when @xmath24 . at low values of @xmath11
the 6.4 kev line dominates , and the ew can not exceed 400 ev .
this turn - around between the ( apparent ) strength of the neutral and ionized lines when the incident spectrum is added in is caused by the effects of absorption by other elements in the gas . when the slab is weakly illuminated much of the flux around 6 kev
is absorbed by other metals in the gas , particularly oxygen , and is re - emitted as softer x - rays .
this results in a large difference in flux between the incidence spectrum and the reflected spectrum at 6 kev .
in contrast , when @xmath20 all the metals except for iron are almost completely ionized , so there is very little absorption around the fe k@xmath0 line .
thus , when the reflection and incident spectra are added together , the he - like 6.7 kev line is more prominent than the neutral line . to summarize , when iron abundances up to 10.1@xmath2 solar were considered , the maximum fe k@xmath0 ew is about 8.5 kev and is found when the gas is neutral and illuminated by @xmath14 power - law . in the case when the power - law is just simply added to the reflection spectrum ( i.e. , a reflection fraction of 1 ) , the ews are much smaller ( @xmath25 kev ) with the largest values being found when the gas is ionized and the 6.7 kev line of he - like fe is dominant . in this case
, it seems to be very difficult to obtain fe k@xmath0 lines with ews greater than 700 ev .
the results presented in the previous section showed that increasing the iron abundance would increase the fe k@xmath0 ew , but its influence seemed to weaken as it grew .
it is interesting to investigate this point since we have only considered abundances up to 10.1@xmath2 solar .
is it possible to obtain an arbitrarily large ew solely by increasing the fe abundance ?
the answer to this question is found in figure [ fig : gamma2-perion ] .
ew / fe abundance ) in the @xmath26-fe abundance plane .
the ews were calculated using the total incident+reflected spectra and we only show the case of @xmath27 .
the contour lines are at 50 , 100 , 200 , 300 , 500 , 700 ev .
the effect of increasing the iron abundance is greatest when it is less than 3@xmath2 solar and when the slab is ionized.,scaledwidth=56.0% ] this plot shows contours of fe k@xmath0 ew normalized by the fe abundance for the case of @xmath27 and a reflection fraction of one . in this case
the contours show a maximum at low fe abundance implying that the most `` power '' out of an fe atom or ion occurs when its abundance is around solar . increasing the amount of iron in the gas
does permit the number of fe k@xmath0 photons emitted to increase , but it also strengthens the line destruction mechanisms of photoabsorption and scattering .
therefore , for a given flux of ionizing photons , competing atomic processes in the gas eventually decreases the efficiency of iron fluorescence .
this is seen particularly when the gas is neutral : the normalized fe k@xmath0 ew falls slowly over a wide range in fe abundance because k@xmath0 photons are destroyed by l shell absorption . at higher values of the ionization parameter
, there is a more rapid drop in the normalized fe k@xmath0 ew .
as is illustrated explicitly in figure [ fig : line - evolution ] , line changes with fe abundance .
total incident+reflection spectra are plotted for the three different values of @xmath1 and three representative values of @xmath11 .
reflection from a slab with a 1.1 , 5.1 and 10.1@xmath2 solar fe abundance is plotted as the solid , short - dashed and long - dashed lines , respectively.,scaledwidth=50.0% ] an increase in the iron abundance from 1.1 to 5.1@xmath2 solar causes a fairly big jump in the line strength , but then only a very small change from 5.1 to 10.1@xmath2 solar . in this case , only k shell absorption is important , but due to the finite number of ionizing photons , the k edge will saturate when the fe abundance is about solar ( from fig . [
fig : gamma2-perion ] ) .
increasing the fe abundance beyond this point will not greatly increase the line emission because there are few photons available to ionize the newly introduced fe atoms .
of course , increasing @xmath11 allows more k shell ionizations to occur , and the normalized ew becomes more uniform with fe abundance .
interestingly , the h - like fe k@xmath0 line at 6.97 kev which is suppressed due to resonance scattering significantly strengthens as the abundance is increased .
these results illustrate that above an fe abundance of @xmath28 5@xmath2 solar there is only a very slow increase in the fe k@xmath0ew .
therefore , it seems unlikely that the ew of the line can be significantly increased by considering iron abundances larger than 10@xmath2 solar .
drb acknowledges financial support from the commonwealth scholarship and fellowship plan and the natural sciences and engineering research council of canada .
acf and rrr acknowledge support from the royal society and the college of the holy cross , respectively .
acf and rrr thank a. young for his work on this project .
antonelli a. , et al . ,
2000 , apj , 545 , l39 bai t. , 1979 , solar phys .
, 62 , 113 ballantyne d.r . ,
ramirez - ruiz e. , 2001 , apj , 559 , l83 ballantyne d.r . ,
ross r.r . ,
2001 , mnras , submitted basko m.m .
, 1978 , apj , 223 , 268 basko m.m .
, sunyaev r.a .
, titarchuk l.g . , 1974 , a&a , 31 , 249 done c. , osborne j.p . , beardmore a.p . , 1995 , mnras , 276 , 483 ebisawa k. , ueda y. , inoue h. , tanaka y. , white n.e . , 1996 , apj , 467 , 419 fabian a.c . , iwasawa k. , reynolds c.s . ,
young a.j . , 2000 , pasp , 112 , 1145 george i.m . , fabian a.c . , 1991 , mnras , 249 , 352 ghisellini g. , haardt f. , matt g. , 1994 , mnras , 267 , 743 matt g. , perola g.c . , piro l. , 1991 , a&a , 247 , 25 matt g. , fabian a.c .
, reynolds c.s . , 1997 , mnras , 289 , 175 matt g. , fabian a.c . , ross r.r . , 1993 , mnras , 262 , 179 matt g. , fabian a.c . , ross r.r . , 1996 , mnras , 280 , 823 morrison r. , mccammon d. , 1983 , apj , 270 , 119 nandra k. , pounds k.a . , 1994 , mnras , 268 , 405 nandra k. , george i.m . , mushotzky r.f . , turner t.j . , yaqoob t. , 1997 , apj , 477 , 602 nayakshin s. , 2000 , apj , 540 , l37 piro l. , et al . , 1999 , apj , 514 , l73 piro l. , et al . , 2000 , science , 290 , 955 pravdo s.h . , 1979 , in ( cospar ) x - ray astronomy , ed .
baity and l.e .
peterson ( oxford : pergamon press ) , p. 169
, mszros , p. , 2000 ,
apj , 545 , l73 reynolds c.s .
, fabian a.c . , makishima k. , fukazawa y. , tamura t. , 1994 , mnras , 268 , l55 ross r.r . , fabian a.c . , 1993 , mnras , 261 , 74 ross r.r . , fabian a.c
, brandt w.n . , 1996 , mnras , 278 , 1082 ross r.r . , fabian a.c .
, young a.j . , 1999 ,
mnras , 306 , 461 swank j.h .
, fabian a.c . , ross r.r . , 1984 ,
apj , 280 , 734 tanaka y. , et al . , 1995 , nature , 344 , 747 van teeseling a. , kaastra j.s . , heise j. , 1996 , a&a , 312 , 186 vietri m. , ghisellini g. , lazzati d. , fiore f. , stella l. , 2001 , apj , 550 , l43 | iron k@xmath0 emission from photoionized and optically thick material is observed in a variety of astrophysical environments including x - ray binaries , active galactic nuclei , and possibly gamma - ray bursts .
this paper presents calculations showing how the equivalent width ( ew ) of the fe k@xmath0 line depends on the iron abundance of the illuminated gas and its ionization state two variables subject to significant cosmic scatter .
reflection spectra from a constant density slab which is illuminated with a power - law spectrum with photon - index @xmath1 are computed using the code of ross & fabian . when the fe k@xmath0 ew is measured from the reflection spectra alone
, we find that it can reach values greater than 6 kev if the fe abundance is about 10@xmath2 solar and the illuminated gas is neutral .
ews of about 1 kev are obtained when the gas is ionized .
in contrast , when the ew is measured from the incident+reflected spectrum , the largest ews are @xmath3 kev and are found when the gas is ionized .
when @xmath1 is increased , the fe k@xmath0 line generally weakens , but significant emission can persist to larger ionization parameters .
the iron abundance has its greatest impact on the ew when it is less than 5@xmath2 solar .
when the abundance is further increased , the line strengthens only marginally .
therefore , we conclude that fe k@xmath0 lines with ews much greater than 800 ev are unlikely to be produced by gas with a supersolar fe abundance
. these results should be useful in interpreting fe k@xmath0 emission whenever it arises from optically thick fluorescence .
[ firstpage ] gamma rays : bursts line : formation line : profiles radiative transfer galaxies : active x - rays : galaxies x - rays : general |
according to the hypothesis of realism , reality is out there , available to us to be experienced .
therefore , what we know about reality comes from our experiences of it , and from the worldviews that we are able to construct when we order these experiences into a possibly consistent map of relations .
no doubts , the human process of creation of a worldview started a long time ago , in pre - cultural and pre - scientific times , when our experiences of reality were only of the `` ordinary kind , '' i.e. , obtained using exclusively our human macroscopic body as a measuring apparatus , plus a few very basic tools . from the ordering of these experiences , a first `` clothing and decoration '' of reality resulted , consisting in the identification of those portions of it that were characterizable by aggregates of sufficiently stable properties , like having a certain amount of weight , a given size , shape , state of movement , orientation , etc .
these `` clusters of properties '' are what we today call _ classical entities _ : the macroscopic objects of our everyday life , and the astronomical objects that we can observe moving in the sky , like the moon and the sun .
we can say that this `` clothing and decoration '' of reality happened ( and happens ) typically in two different `` directions of penetration . ''
the first one is a _ penetration in depth _ , which precisely corresponds to the process we have just mentioned of identifying the entities forming `` stable aggregates of elements of reality . ''
the second one is a _ penetration in width _ , corresponding to a process of organization of the relations existing between the different entities , i.e. , between those aggregates of properties that appear to be separate @xcite ( not influencing each other in a significant way ) .
this penetration in width of our reality can be considered as an ordering process giving rise to a _ space _ , and more specifically to our _ euclidean three - dimensional space _ , which therefore should be considered , first of all , a _ space of relations_. according to the above
, it is clear that when a given ensemble of experiences is properly ordered and organized , a specific _ theater of reality _ will emerge , inside of which a given typology of actors can perform and relate , in predetermined ways . our euclidean physical space , in that sense ,
should not be understood as an external `` container , '' accommodating the play of these classical actors , but really as the manifestation of the structure of their possible relations .
it is then also clear that when an ensemble of new experiences becomes all of a sudden available , our first attempt will be that of trying to find a place for them in the theater of reality that we have already constructed .
this also because , as time passes by , it is easy to forget about the process of its construction and start believing that all of reality should necessarily fit into it , as the theater and its content , and reality , would just be one and the same thing .
as is known , an amazing ensemble of new experiences that came shaking the structure of our classical theater are those that were obtained in the investigation of the microscopic layer of our reality , in confirmation of that corpus of knowledge known as _ quantum mechanics_. examples are the famous experiments of _ stern _ and _ gerlach _ with silver atoms @xcite , of _ rauch et al . _ with neutrons @xcite , and of _ aspect et al . _ with entangled photons @xcite .
( another important ensemble of experiments that have undermined the structure of our classical theater are those that have led to the discovery of _ relativity theory _ , but we shall not be concerned by them in this article ) .
thanks to these very sophisticated experiments , we could access elements of reality that in the past were totally beyond our reach , because of our too coarse physical senses , but also because our natural environment did not provide the controlled conditions of a modern laboratory .
these experiments revealed to us that although these new elements of reality were in part similar to those we had so far experienced , at the same time they were sufficiently different to spoil our efforts to incorporate them in our classical _
euclidean theater_. a typical example is that of spins , which we will discuss more particularly in this article .
a spin is usually described as an intrinsic angular momentum carried by a microscopic entity , like an electron , a proton , a neutron , or by a combination of elementary entities ( composite entity ) , like an atomic nuclei . however , although it possesses the same physical dimension of an angular momentum carried by a macroscopic body ( the dimension of the planck constant @xmath0 ) , it can not be associated with any specific rotation . for instance , because if the spin is calculated on a classical basis , as a rotation in space , such a rotational movement would yield a superluminal velocity along the microscopic particles periphery @xcite , in violation of the relativistic limit . also ,
in the case of fractionary spins , we know that a @xmath1 rotation will not bring the spin entity back in the same state @xcite , as is instead the case for the actors of the classical theater .
this impossibility of understanding a spin as a state of rotation in the three - dimensional space can be further evidenced by observing that it can not generally and consistently be represented as a three - dimensional vector , as is the case for the angular momentum of a macroscopic object .
this is so because in quantum mechanics a spin ( and more generally an angular momentum ) is described by an operator , and therefore can not be simply drawn as a three - dimensional vector quantity whose components would be real numbers . to see this more explicitly ,
consider a general spin-@xmath2 entity , with @xmath3 .
its states are described by complex valued vectors in a @xmath4-dimensional hilbert space @xmath5 , with @xmath6 .
the spin observable @xmath7 , along the space direction @xmath8 , is then given by the product @xmath9 , where @xmath10 is a unit vector : @xmath11 , and the @xmath12 , @xmath13 , are the three orthonormal vectors forming the canonical basis of the euclidean space .
the three observables @xmath14 , @xmath13 , are @xmath15 self - adjoint matrices obeying the commutation relations : @xmath16=i\hbar s_3 $ ] , @xmath17=i\hbar s_1 $ ] and @xmath18=i\hbar s_2 $ ] , from which one can also deduce that @xmath19 ( this is only true for non - composite spin entities ) , with @xmath20 the identity matrix , so that all spin states are also eigenstates of @xmath21 , with @xmath4 times degenerate eigenvalue @xmath22 .
we also denote by @xmath23 , @xmath24 , the @xmath4 eignevectors of the spin components @xmath25 , @xmath26 , which form a basis of @xmath27 : @xmath28 , @xmath29 , @xmath13 .
when a spin-@xmath2 entity is , say , in the eigenstate @xmath30 of @xmath31 , we can try to classically describe the situation by drawing a vector of length @xmath32 ( according to @xmath33 ) , whose projection onto @xmath34 is precisely @xmath35 ( according to @xmath36 ) .
in fact , there is an entire collection of possible vectors of length @xmath32 , whose projections onto @xmath34 give the value @xmath35 : they form a cone of height @xmath37 , lateral height @xmath32 , and radius @xmath38 ( see fig . [ spin3d ] , for the @xmath39 case ) .
the vectors belonging to the cone , however , can not consistently represent the spin of a quantum entity in the state @xmath30 .
indeed , we know that if we measure the spin along any other direction , all the values @xmath40 are in principle obtainable .
this , however , is incompatible with the geometry of the cone . to see this ,
let us just consider for simplicity the spin-@xmath41 case .
if we take one of the cone s vectors as the direction of a spin measurement ( here understood in the classical sense ) , then by orthogonally projecting all the cone s vectors along that specific direction , we obtain the interval of possible values @xmath42 $ ] ( see fig .
[ spin3d ] ) , which is manifestly incompatible with the experimental data , as it does nt contain the eigenvalue @xmath43 .
the above constitutes a very simple no - go argument showing that it is impossible to directly associate a quantum spin eigenstate ( and a fortiori , of course , a superposition of spin eigenstates ) with an angluar momentum vector having a specific direction in space , not even if instead of a single direction we accept to associate to it an entire collection of indeterminate vectors lying on a cone .
this means that not only a spin eigenstate can not be understood as an entity belonging to our three - dimensional spatial theater , it neither can be understood as a higher - dimensional classical - like entity characterized by multiple directions , that would simply `` cast a three - dimensional shadow '' onto it ( in a sort of view la _
edwin a. abbott _ ) .
so , coming back to our initial discussion , we see that through our quantum experiments we have recently ( in the time scale of human evolution ) opened a window to an entire new sector of our reality , which requires us to construct a brand new theater , to find a suitable ( non - spatial ) `` place '' for the new quantum actors , in which their show can be represented .
this of course will be quite a non - ordinary theater according to our classical criteria , as these quantum actors can behave in a way that is technically impossible to the classical ones .
but considering the existence of these two theaters , the quantum and the classical , two questions arise .
the first question is about the nature of their interactions . indeed , quantum entities , although not in space ( and possibly also not in time )
, are nevertheless available to leave traces into it .
if this would not be the case , we would never have learned about their existence , as our measuring apparatuses are macroscopic objects , extensions of our human bodies , which means that we can only explore the content of the quantum theater as from the euclidean one .
in quantum mechanics the effects of these interactions are described by the born rule and the associated projection postulate , so the question is : _ how does a quantum entity interact with a classical one ( a measuring apparatus ) during a measurement process , so as to produce the different possible outcomes , in accordance with the predictions of the born rule ? considering that a measurement process can not be understood as a mere process of discovery of a shadow that a quantum entity would cast onto the euclidean stage , how can we describe the process through which it leaves a detectable trace onto it ? _ the second question is about finding what are the possible relations between the elements of reality characterizing the entities living in the quantum ( non - spatial ) theater , and those characterizing the classical entities ( spatial objects ) located in our human classical theater . in other terms
, it is about finding the possible connections between what our sophisticated experimental apparatuses indicate to us to be real , and what we know ( or think ) to be real from our direct pre - scientific human experience
. a more specific formulation of this question , in relation to spin entities , would be the following : _ what is the general relation between the non - spatial spin eigenstates and the space directions in the euclidean space , and is it possible to have a precise geometrical description of such relation ? _ the purpose of this article is to offer an answer to the above two questions . for this
, the article is organized as follows . in sec .
[ the bloch sphere ] , we introduce the standard bloch sphere representation , recalling that for spin-@xmath41 entities there is a direct ( one - to - one ) correspondence between spin states and space directions ( each state being oriented along a different space direction ) . in sec . [ the extended bloch sphere ] , we show how to extend the bloch result to also include the measurements , and in sec .
[ general measurement ] we explain how the model can be naturally generalized to describe entities of arbitrary dimension @xmath4 , answering in this way the first question .
then , in sec . [ spin and space ]
, we use the extended bloch representation to show that the spin eigenstates , beyond the spin-@xmath41 case , are not anymore oriented along space directions , and therefore should be considered to be genuine non - spatial elements of our reality ( not belonging to our classical theater ) .
however , we also show that the different eigenstates are oriented in a very specific way with respect to the space directions describing the measurements for which they are eigenstates , within the extended bloch sphere ( the quantum theater ) , answering in this way the second question . in sec .
[ composition of spins ] , we show that this relation between spin eigenstates and space directions remains valid also for composite spin entities , and in sec .
[ connecting elements of reality ] we explain how the correspondence between the elements of reality in our classical theater , and those in the quantum one , should be understood , and what are the limitations with regard to spins . finally , in sec .
[ conclusion ] , we introduce the notion of _ multiplex realism _ and offer some concluding remarks .
in the previous section we have shown that the representation of quantum spin eigenstates of different magnitudes by means of cones of vectors in a 3-dimensional space is inconsistent , and therefore misleading , as it conveys the wrong idea that a spin eigenstate could be put into full correspondence with an intrinsic classical angular momentum , although of an indeterminate nature .
on the other hand , if we renounce associating spin eigenstates with classical angular momenta , it is certainly possible , at least in the special case of spin-@xmath41 entities , to characterize each state by means of a specific direction of space .
however , for spin entities of magnitude @xmath44 , this will no longer be possible , as to characterize the different eigenstates , a @xmath45-dimensional quantum theater will be required . but let us start by considering the simple spin-@xmath41 situation .
let @xmath46 be the two eigenvectors of the spin operator @xmath7 , oriented along an arbitrary direction @xmath10 , for the eigenvalues @xmath47 . in the following , for simplicity
, we will simply write : @xmath48
. then , also the vector - states @xmath49 , with @xmath50 , are eigenvectors of @xmath7 , for the same eigenvalues . in other terms , the vectors @xmath51 and @xmath49 represent the same physical states , in accordance with the known fact that no measurement can determine a global phase factor .
the reason for this is that a measurement is a process that is precisely designed to single out a given entity , in order to specifically measure some of its properties .
this means that a measurement is a context conceived to only act on the entity which is meant to be measured , and not , say , on combinations of that entity with other entities .
this isolation of the measured entity in the measurement context thus excludes the possibility that a global phase factor would be able to produce interference effects , and therefore become observable .
a way to make fully explicit this isolation of an entity in a measurement context , causing the global phase of its vector - state @xmath52 to become irrelevant , is to describe its state by means of a one - dimensional projection operator @xmath53 .
indeed , knowing @xmath54 , one can always reconstruct @xmath52 , but only up to a global phase factor . to show this ,
we write @xmath55 , so that ( to simplify the notation , we set : @xmath56 ) : @xmath57 .
we assume that @xmath58 , and we fix the global phase of @xmath52 in a way that @xmath59 is a real strictly positive number .
then , since @xmath60 , we have @xmath61 .
also , @xmath62 , so that @xmath63 . and of course , if @xmath64 , then @xmath65 is simply a phase factor .
so , when we go from the _ vector - state _
representation @xmath52 , to the associated ( rank-1 projection ) _ operator - state _
representation @xmath54 , the only information we loose is that relative to the vector s global phase , which plays no role in a measurement .
therefore , the operator - state @xmath54 can be considered to be the proper mathematical object representing the state in which the spin entity ( and more generally any quantum entity ) is prepared and isolated in view of being subjected to a measurement process .
our next step is to show how to represent the spin states @xmath54 as directions in space .
this is a well - known representation that dates back to 1892 , when the french physicist and mathematician _ henri poincar _ discovered that a surprisingly simple representation of the polarization states of electromagnetic radiation could be obtained by representing the polarization ellipse on a complex plane , and then further projecting such plane onto a sphere @xcite . in 1946 , this representation was adapted by the swiss physicist _
felix bloch _
@xcite to represent the states of two - level quantum systems , like spin-@xmath41 entities , in what is today known as the _ poincar sphere _ or _
bloch sphere_. this representation is a consequence of the observation that a general complex @xmath66 matrix can always be written as a linear combination of four matrices @xmath67 , @xmath68 , where @xmath69 , is the identity matrix , and @xmath70 , @xmath71 , are the three pauli matrices . in the eigenbasis of @xmath72 , these four matrices
take the explicit form : @xmath73,\quad \sigma_1= \left [ \begin{array}{cc } 0 & 1 \\ 1 & 0 \end{array } \right],\quad \sigma_2= \left[\begin{array}{cc } 0 & -i \\ i & 0 \end{array } \right],\quad \sigma_3= \left [ \begin{array}{cc } 1 & 0 \\ 0 & -1 \end{array } \right ] .
\label{pauli - matrices}\ ] ] it is then straightforward to show that they are all mutually orthogonal , in the sense that @xmath74 , from which it follows that ( setting @xmath75 ) : @xmath76 , @xmath71 . therefore , the state @xmath54 can generally be written as the linear combination of the four matrices @xmath67 : @xmath77 , where the factor @xmath41 has been introduced for convenience .
since @xmath54 and the @xmath67 are all self - adjoint , the coefficients @xmath78 must be real . to determine them , we observe that @xmath79 , from which it follows that @xmath80 , since @xmath81 .
considering that @xmath82 , we have @xmath83 , implying that @xmath84 .
we also know that @xmath54 obeys the eigenvalue equation @xmath85 , which we can write as @xmath86 .
taking the trace of this expression , we thus find that : @xmath87 , from which we conclude that @xmath88 , @xmath13 .
this means that @xmath54 is fully determined by the unit real vector @xmath10 , and we can write in compact form : @xmath89 where @xmath90 .
in particular , if we write the unit vector @xmath10 in spherical coordinates , that is : @xmath91 , the operator - state @xmath92 takes the explicit form : @xmath93 = \left [ \begin{array}{cc } \cos^2{\theta\over 2 } & \sin{\theta\over 2 } \cos{\theta\over 2}e^{-i\phi } \\
\sin{\theta\over 2 } \cos{\theta\over 2 } e^{i\phi } & \sin^2{\theta\over 2 } \end{array } \right ] .
\label{spherical - coordinates}\ ] ] the spin eigenstates of a spin-@xmath41 entity can thus be fully determined by specifying the unit vectors @xmath10 defining the space direction for which the associated spin observable @xmath7 , if measured , would produce with certainty the outcome @xmath94 .
this representation ( see fig . [ bloch - sphere ] ) is however very different from the ( incomplete ) cone representation described in the previous section , as is clear that none of the classical spin vectors lying on the cone point to the same direction as the unit vector representative of the state .
apart from that difference , the bloch sphere allows us to maintain a one - to - one correspondence between spin states and space directions , and is valid for arbitrary measurements .
however , such direct correspondence will be lost when considering spin entities of magnitude greater than @xmath41 , in the sense that the directions of space will not anymore be sufficient to characterize all the possible relations between states and measurements , which can only be represented in an extended space of higher dimensionality . however
, each spin eigenstate will maintain a very specific relation ( orientation ) with respect to the space direction characterizing its measurement .
but before showing this , we need to extend the bloch representation .
in the previous section we have described the well - known bloch sphere representation for spin-@xmath41 entities ( also valid for general qubit systems ) .
this is a representation that works for the states , and not for the measurements .
however , it is possible to extend it to also include a modelization of the different possible measurements , as was shown by one of us forty years after the work of bloch , providing at the same time a plausible explanation of the origin of the born rule @xcite . before showing in this section how this can be done , it is worth mentioning that the perspective we are taking here is that which consists in taking the quantum formalism very seriously . by this
we mean that we consider that a hilbertian vector - state , or ( when its global phase becomes irrelevant ) its associated operator - state , does actually describe the real state of the physical entity under investigation , and not just our knowledge or our beliefs about its condition .
accordingly , we also consider that a measurement process is an objective physical process , bringing the entity from an initial pre - measurement state to a final post - measurement state , in a way that can not be determined in advance .
however , what can be determined in advance are the probabilities of the different transitions , producing the different possible outcomes of the measurement .
once the quantum formalism is taken seriously in the above sense , we have of course to face the challenge of understanding what happens during the quantum measurement process .
indeed , as is well known , measurements are not explained in the standard quantum formalism , as the born rule of correspondence , with which the different probabilities are calculated , is just postulated . to be more specific , consider a spin-@xmath41 entity prepared in state @xmath54 , subjected to the measurement of the observable @xmath72 . according to the projection postulate ,
two transitions can possibly occur during the measurement process : the transition @xmath95 , associated with the eigenvalue @xmath94 , or the transition @xmath96 , associated with the eigenvalue @xmath43 .
the born rule then tells us that the probabilities associated with these two transitions are : @xmath97 where for the third equality we have used the fact that the pauli matrices have zero trace , and for the last one that they are mutually orthogonal .
it is possible to write the above expressions in a more compact form by introducing the polar angle @xmath98 between @xmath10 and @xmath34 , given by @xmath99 ( see fig .
[ bloch - sphere ] ) .
this gives : @xmath100 if the states @xmath54 are assumed to express an objective condition of the physical entity , it is clear that a measurement has also to be understood not just as a process of _ discovery _ , but also as a process of _ creation_. the discovery aspect is related to the fact that the statistics of outcomes depends on the pre - measurement state ( which for instance can be recovered by quantum tomography techniques ) ; the creation aspect , on the other hand , is related to the fact that the process literally creates the property that is observed , in a non - predictable way ( apart of course in the special situation where the pre - measurement state is an eigenstate of the measurement ) .
what we will now show is how the measurement of an observable , for example the spin observable @xmath72 , producing the probabilities ( [ transition - probabilities2 ] ) , can be modeled within the bloch sphere , in what is called the _ extended bloch representation _ @xcite . to do so
, we will need more ( pure ) states than those that are usually considered in the standard formalism . indeed
, to describe the indeterministic dynamics characterizing a quantum measurement , and derive from it the born rule , we need to go inside the bloch sphere , i.e. , to also consider non - unit representative vectors .
we start by specifying to what kind of operators the non - unit vectors @xmath101 , living inside the sphere , are associated with .
let @xmath102 .
clearly , even if @xmath103 , we still have @xmath104 .
also , considering that the pauli matrices have eigenvalues @xmath105 , it follows that @xmath106 has eigenvalues @xmath107 , and therefore is a positive semidefinite operator .
being also a manifestly self - adjoint operator ( @xmath101 is a real vector ) , it corresponds to what is usually called a _ density matrix _ and is generally interpreted as a classical statistical mixture of pure states .
however , such interpretation is not without difficulties , as a same density matrix can have arbitrarily many different representations as a mixture of one - dimensional projection operators @xcite .
our assumption here is that density matrices are able to describe not only classical mixtures of states , but also ( extended ) pure states , and precisely those pure states that describe how an entity , during a measurement process , approaches the measurement s `` region of potentiality , '' triggering in this way the actualization of a specific interaction . since the bloch sphere is a sphere of states , to geometrically represent measurements inside of it we have to consider their associated eigenstates . in the case of the observable @xmath72
, we have the two eigenstates @xmath108 and @xmath109 , represented in the sphere by the two zenith and nadir points @xmath34 and @xmath110 , respectively .
in addition to these two antipodal points , we consider all the intermediary points forming the one - dimensional segment of length 2 going from @xmath110 to @xmath34 , corresponding to the vertical diameter of the sphere ( see fig . [ bloch - sphere2 ] ) .
this line segment , or diameter , associated with @xmath72 , precisely corresponds to that _ potentiality region _ characterizing the measurement context , which is responsible for the indeterministic `` collapse '' ( as we shall see in a moment ) . but
in order for this to occur , the entity has first to enter into contact with that region .
this means that the measurement process has to involve a preparatory _
phase , through which the pre - measurement state @xmath54 is brought into contact with the latter .
this process corresponds to the immersion of a point particle associated with the vector @xmath10 , representative of the entity s pre - measurement state , from the surface of the sphere to its interior , to reach the line segment representing the potentiality region , along an orthogonal path ( see fig .
[ bloch - sphere2 ] ) .
if we denote @xmath111 the point on the line segment obtained in this way , that is , @xmath112 , we can describe this deterministic movement of approach of the potentiality region by means of a parameter @xmath113 , which is varied from 0 to 1 : @xmath114 .
clearly , @xmath115 is the initial condition , and @xmath116 is the final condition on the potentiality region . before describing the purely indeterministic part of the measurement , it is instructive to write the generalized operator - state @xmath117 in explicit terms . with @xmath91 and @xmath118
, we have @xmath119 , so that @xmath120 takes the explicit form : @xmath121 . \label{decoherence}\ ] ] as it can be seen in this explicit expression , the deterministic approach of the entity s state towards the potentiality region corresponds to a decoherence - like process , causing the off - diagonal elements of the operator - state @xmath122 to gradually vanish , as @xmath123 , so that the `` on - potentiality region '' state has the form of a fully reduced density operator : @xmath124 .
we now describe the second phase of the measurement process , corresponding to that purely indeterministic dynamics which is responsible for the emergence of the quantum probabilities , in accordance with the born rule . for this
, we have to think of the potentiality region as a region made of a uniform substance which is not only _ attractive _ ( as it causes the initial state to be orthogonally attracted towards it ) , but also _ unstable _ and _ elastic_. a simple image we can use is that of a uniform elastic band that would have been stretched between the two anchor points @xmath110 and @xmath34 , with the state of the entity represented by a point particle stuck onto it , at point @xmath111 .
the instability of the substance means that the elastic at some moment will break , at some unpredictable point , so causing its splitting into two halves , which will then contract towards the respective anchor points . depending
on which of these two halves the point particle is attached to , it will either be drawn to point @xmath110 , or to point @xmath34 ( see fig .
[ bloch - sphere3 ] ) .
more precisely , let @xmath125 be the line segment between @xmath111 and @xmath110 , and @xmath126 be the line segment between @xmath34 and @xmath111 .
their lengths ( lebesgue measures ) are : @xmath127 considering that the first immersive phase of the measurement process is perfectly deterministic , it is clear that the transition probability @xmath128 is nothing but the probability that the point particle is drawn to point @xmath129 , which in turn corresponds to the probability @xmath130 that the disintegration point belongs to region @xmath131 .
being the elastic substance , by hypothesis , uniform , and of total length @xmath132 , we thus have : @xmath133 which are precisely the quantum mechanical probabilities ( [ transition - probabilities2 ] ) predicted by the born rule . in other terms , the standard bloch sphere representation can be extended to also include a description of the different possible measurements , in accordance with the predictions of the born rule . before explaining in the next section how this representation can be further generalized to an arbitrary number of dimensions , a few remarks are in order .
what we have described is clearly a measurement of the first kind . indeed , once the point particle has reached one of the two outcome positions
@xmath129 , if subjected again to the same measurement , being already located in one of the two anchor points of the elastic , we have @xmath134 , so that its position can not be further changed by its collapse .
the disintegration points can be interpreted as variables specifying the _ measurement interactions_. thus , the model provides a consistent hidden - measurement interpretation of the quantum probabilities , as epistemic quantities characterizing our lack of knowledge regarding the measurement interaction that is actualized between the measured entity and the measuring apparatus , at each run of the experiment .
it is worth observing that almost each measurement interaction gives rise to a purely deterministic process , changing the state of the entity from @xmath111 to either @xmath34 or @xmath110 , depending whether @xmath135 , or @xmath136 .
we say ` almost ' because when @xmath137 , that is , when coincides with the point separating region @xmath126 from region @xmath125 , we have a situation of classical unstable equilibrium .
this point of classical instability is at the origin of the distinction between the two outcomes , but being of measure zero , it does nt contribute to the values of the probabilities associated with them .
in other terms , although the border points @xmath137 are the `` source of the possibilities , '' they do not contribute to the values of the probabilities that are associated with them .
the preliminary 1986 study that allowed to extend the 3-dimensional bloch sphere representation to also include measurements ( as shown in the previous section ) generated over the years a number of works , further exploring the explicative power contained in this modelization , also called the _ hidden - measurement interpretation _ of quantum mechanics ( see @xcite and the references cited therein ) . according to it ,
a quantum measurement is an experimental context characterized by a lack of knowledge not about the state of the measured entity , but about the interaction between the measuring apparatus and the measured entity . and it is precisely through a mechanism of _ actualization of potential interactions _ ( or of spatialization of non - spatial interactions ) that the non - ordinary entities belonging to the quantum theater would leave their ephemeral traces into our classical spatial theater ( the first question we addressed in the introduction ) . for some time
this hidden - measurement extension of the bloch sphere , called the _ @xmath138-model _ ( which also allowed an exploration of the intermediary region between classical and quantum ) was taken in serious consideration only by a small number of physicists working in the foundations of physical theories .
this probably because of the existence of the so - called no - go theorems , like those of gleason @xcite and kochen - specker @xcite , which were known to be valid only for @xmath4-dimensional hilbert spaces with @xmath139 , hence not for the special case of two - level systems ( qubit ) , which therefore were considered to be pathological .
accordingly , the extended bloch representation was mostly considered as a mathematical curiosity , but this was a misconception , as the hidden - measurement interpretation has little to do with a classical hidden - variable theory , as is clear that the lack of knowledge is not associated with the state of the measured entity , but with its measurement - interactions ( so that the no - go theorems do not apply ) . despite this prejudice that no hidden - measurement modelizations could be given for dimensions @xmath139 , new results became available over the years , showing that the hidden - measurement mechanism was by no means restricted to two - dimensional situations @xcite .
these promising results were however not totally convincing , as what was still lacking was a natural generalization of the bloch sphere representation beyond the two - dimensional situation .
things changed in more recent times , when some important mathematical results became available , precisely providing this much - sought generalized bloch sphere representation , which exploits the properties of the so - called generators of @xmath140 , the special unitary group of degree @xmath4 @xcite .
thanks to these results , very recently we could extend the generalized bloch construction to also include a full description of the measurements ( including the degenerate ones ) , offering in this way what we think is a general and convincing solution to the measurement problem @xcite . in this section
we will explain how a @xmath4-outcome measurement can precisely be described , and therefore explained , in this extended bloch modelization .
in other terms , we will now consider an entity whose hilbert state is @xmath5 ( if it is a spin-@xmath2 entity , then @xmath141 ) .
what we need to observe is that , as already mentioned , the 3-dimensional bloch - sphere representation is based on the property of the pauli matrices and the identity operator of generating a basis for all linear operators on @xmath142 , so allowing to expand any operator - state on them , in the form : @xmath143 .
this correspondence between @xmath66 operator - states and real vectors in the three - dimensional unit ball is the expression of the well - known homomorphism between @xmath144 and @xmath145 .
such homomorphism can not be extended beyond the @xmath146 situation , but one can still represent operator - states as real vectors in a unit sphere , which however will not be anymore three - dimensional , nor completely filled with states . more precisely , the reason why operator - states can still be represented in this way is that one can always find @xmath147 self - adjoint @xmath15 matrices @xmath148 , @xmath149 , of zero trace , which together with the identity operator form a basis for all linear operators acting on @xmath150 .
these matrices are the so - called generators of @xmath140 , and if we choose the normalization @xmath151 , a convenient construction is @xcite : @xmath152 , with : @xmath153 where @xmath154 is an arbitrary orthonormal basis of @xmath155 .
for the two - dimensional case ( @xmath146 ) , the generators are @xmath156 , and if @xmath157 is taken to be the canonical basis , they precisely correspond to the pauli matrices ( [ pauli - matrices ] ) . for the three - dimensional case ( @xmath158 ) ,
the generators are @xmath159 , and correspond to the so - called gell - mann matrices .
so , a general operator - state can always be written as the linear combination : @xmath160 where for convenience we have introduced the constant @xmath161 . taking the trace of @xmath162 , then using the orthogonality of the @xmath148 , after a simple calculation one finds that : @xmath163 , and since an operator - state generally obeys @xmath164 , it follows that @xmath165 , that is , that the representative vectors @xmath101 belong to a @xmath166-dimensional unit ball @xmath167 . clearly , @xmath168 iff @xmath169 , i.e. , if @xmath106 is a one - dimensional projection operator ( a pure state , in the standard quantum terminology ) . similarly to the @xmath146 situation
, states can therefore be represented as vectors living in a unit ball . on its surface ,
one finds the rank - one projection operators , and in its interior the density - matrices , i.e. , the operator - states .
the important difference with the standard bloch representation is that the ball is now @xmath166-dimensional , i.e. , it is not anymore representable in @xmath170 , and , more importantly , only a small portion of it contains states . indeed ,
if it is true that for every operator - state @xmath171 one can always find a vector @xmath172 such that @xmath171 can be written in the form ( [ formulanxn ] ) , the converse will not be generally true : given a vector @xmath172 , the operator ( [ formulanxn ] ) does not necessarily describe a state .
the reason for this is that @xmath106 , to be a state , needs not only to be of unit trace and self - adjoint , but also positive semidefinite , which is not automatically guaranteed for an operator written as the real linear combination ( [ formulanxn ] ) .
just to give an example , consider the unit vector @xmath173 .
then @xmath174 , and according to ( [ generatorsw ] ) @xmath175 , so that @xmath176 , which for @xmath177 is clearly a matrix with a strictly negative eigenvalue @xmath178 , and therefore can not be positive semidefinite .
what is important to observe is that although @xmath167 is only partially filled with states , and that the shape of the region containing the states is rather complex , it is however a _ closed convex _ region .
this follows from the well - known fact that a convex linear combination of operator - states is again an operator - state .
thus , if @xmath179 , with @xmath180 , @xmath181 , and @xmath182 and @xmath183 are representative of two operator - states , from ( [ formulanxn ] ) we immediately obtain that @xmath184 is again an operator - state , being a convex linear combination of operator states .
we thus conclude that a vector @xmath101 , which is a convex linear combination of two vectors @xmath182 and @xmath183 , representative of states , is also representative of a bona fide state @xmath106 .
we now explain how measurements can be represented within @xmath167 , so generalizing the previous two - dimensional measurement model .
we start by observing that also in the general @xmath4-dimensional situation a vector - state @xmath185 can always be represented , in a measurement context , by a one - dimensional projection operator @xmath186 , as the only information one looses in the passage from @xmath185 to @xmath187 is the global phase of the former . to show this , we write : @xmath188 and @xmath189 , so that @xmath190 . for @xmath191
, we have @xmath192 . assuming that @xmath193 ( if this is not the case we can reason with @xmath194 , and so on ) , we can always fix the global phase of @xmath185 so as to have @xmath195 . then @xmath196 , and we obtain : @xmath197 , @xmath198 . thus , apart a global phase factor , we can fully reconstruct @xmath185 from @xmath187 .
our goal is to explain how the values of the transition probabilities @xmath199 can be derived by means of a hidden - measurement mechanism , where @xmath200 , and the orthonormal vectors @xmath201 are the eigenvectors of the spectral decomposition of an arbitrary observable @xmath202 , which for the moment we shall assume to be non - degenerate ( the eigenvalues @xmath78 are all distinct ) .
let @xmath10 be the unit vector representative of @xmath203 , and @xmath204 the unit vectors representative of @xmath205 , @xmath206 .
we then have : @xmath207 = { 1\over n } + { c_n^2\over n^2}\,{\rm tr}\,({\bf n } \cdot\mbox{\boldmath$\lambda$})\ , ( { \bf n}_i\cdot\mbox{\boldmath$\lambda$})\nonumber\\ & = & { 1\over n } \left[1 + ( n-1)\,{\bf n}\cdot { \bf n}_i\right]={1\over n } \left[1 + ( n-1)\,\cos\theta\right ] , \label{transitiongeneralnxn}\end{aligned}\ ] ] where @xmath208 is the angle between @xmath10 and @xmath204 .
the above formula is clearly a generalization of ( [ transition - probabilities2 ] ) , and we can use it to gain some knowledge on how the observables are represented within the unit ball @xmath167 .
if the initial state is an eigenstate @xmath209 , then we know that @xmath210 .
it follows from ( [ transitiongeneralnxn ] ) that @xmath211 , that is : @xmath212 , for all @xmath213 .
this means that the @xmath4 unit vectors @xmath204 , representative of the eigenstates @xmath214 , @xmath215 , are the vertices of a @xmath216-dimensional _ simplex _ @xmath217 , inscribed in the unit ball , with edges of length @xmath218 .
also , considering that @xmath217 is a convex set of vectors , it immediately follows that all points contained in it are representative of operator - states , in accordance with the fact that the states in @xmath219 form a closed convex subset . for @xmath146 ,
@xmath220 , and @xmath221 is a line segment of length @xmath222 inscribed in a @xmath223-dimensional ball , as we have seen in the previous section . for @xmath158 , @xmath224 , and @xmath225
is an equilateral triangle of area @xmath226 . for @xmath227 , @xmath228 , and @xmath229
is a tetrahedron of volume @xmath230 . for @xmath231 , @xmath232 , and @xmath233
is a pentachoron ; and so on
. we can write ( [ transitiongeneralnxn ] ) in a more compact and simple form by expressing @xmath10 as the sum : @xmath234 , where @xmath111 is the vector obtained by orthogonally projecting @xmath10 onto @xmath217 . since by definition @xmath235 , @xmath236 , ( [ transitiongeneralnxn ] ) becomes @xmath237 $ ] .
also , as @xmath238 , by definition of a simplex it can be uniquely written as a convex linear combination of the @xmath4 vertex vectors @xmath204 : @xmath239 since @xmath240 , for @xmath213 , we have : @xmath241 , and therefore ( [ transitiongeneralnxn ] ) simply becomes : @xmath242 in the @xmath146 case we have seen that the measurement results from the interaction between an abstract point particle representative of the state of the entity and an attractive , elastic and unstable substance uniformly filling the one - dimensional simplex @xmath221 , representative of the measurement context . in the same way ,
the measurement context associated with a general @xmath4-dimensional observable consists of a @xmath216-dimensional simplex @xmath217 , uniformly filled with an attractive , elastic and unstable substance , with the point particle representative of the state that first orthogonally `` falls '' onto it and then is drawn to one of its apex points in an unpredictable way , as a result of the disintegration and collapse of said substance . to see how all this works
, we will only describe here , for simplicity , the @xmath158 situation , as the general situation proceeds according to the same logic and is therefore a straightforward generalization @xcite .
so , the measurement context is in this case represented by a @xmath222-dimensional triangular elastic membrane inscribed in a 8-dimensional ball , and the three possible outcomes are the eigenstates associated with its three vertex vectors @xmath243 , @xmath244 and @xmath245 . if the initial , pre - measurement state @xmath54 is associated with a unit vector @xmath246 , the entity proceeds first with a deterministic movement @xmath247 , @xmath248 $ ] , which brings the state of the entity from the point at the surface @xmath115 , to the on - membrane point @xmath249 , along a path that is orthogonal to @xmath225 ( see fig .
[ spinmachine3d ] ) .
once the particle has reached its on - membrane position @xmath111 , it defines three different triangular subregions @xmath250 , @xmath251 and @xmath252 , delineated by the line segments connecting the particle s position with the three vertex points ( see fig .
[ spinmachine3d ] ) .
one should think of these line segments as `` tension lines '' altering the functioning of the membrane , in the sense of making it less easy to disintegrate along them .
then , at some moment , the membrane disintegrates , at some unpredictable point , belonging to one of these three subregions .
the disintegration then propagates inside that specific subregion , but not into the other two subregions , because of the presence of the tension lines .
this causes the two anchor points of the disintegrating subregion to tear away , producing the detachment of the membrane , which being elastic contracts towards the only remaining anchor point , drawing to that position also the particle that is attached to it , which in this way reaches its final destination ( state ) , corresponding to the outcome of the measurement ( see fig . [ spinmachine3d ] ) .
reasoning in the same way as we did in the @xmath146 case , it is clear that the transition probability @xmath253 is given by the probability @xmath254 that the disintegration point belongs to region @xmath255 . considering that @xmath225 is an equilateral triangle of area @xmath226 , we thus have @xmath256 .
let us consider for instance @xmath250 .
it is a triangle with vertices @xmath244 , @xmath245 and @xmath257 . using the explicit coordinates of the three vertices of @xmath250
, we can easily calculate its area .
for this , we can use a system of coordinate directly in the plane of the triangle , such that @xmath258 , @xmath259 , and @xmath260 . to calculate the area
we can then use the general formula : @xmath261 . after a calculation without difficulties
one finds , using ( [ n - parallelexpansion ] ) : @xmath262 . doing similar calculation for @xmath251 and @xmath252 ,
one obtains that @xmath263 , @xmath13 , so that @xmath264 , @xmath13 , in accordance with the predictions of the quantum mechanical born rule ( [ trans - general ] ) . the hidden - measurement membrane mechanism that we have described for the @xmath146 case in the previous section , and for the @xmath158 case in the present section , generalizes in a natural way to an arbitrary number of dimensions @xmath4 , and we refer the reader to @xcite for all the mathematical details .
the membrane mechanism can also be used to describe measurements of degenerate observables .
then , the subregions associated with the degenerate eigenvalues are fused together and form bigger composite subregions , so that when the initial disintegration point takes place inside one of them , the process draws the particle not to a vertex point of @xmath217 , but to one of its sub - simplexes .
the collapse of the elastic substance remains however compatible with the predictions of the _ lders - von neumann projection formula _ , but to complete the process the particle also has to re - emerge from the sub - simplex region , to deterministically reach its final position , at the surface of the unit ball .
in other words , in the general situation a measurement is to be understood as a tripartite process formed by ( 1 ) an initial deterministic _ decoherence - like _ process , corresponding to the particle reaching the `` on - membrane region of potentiality ; '' ( 2 ) a subsequent indeterministic _ collapse - like _ process , corresponding to the disintegration of the elastic substance filling the simplex , with the particle being drawn to some of its peripheral points ; and ( 3 ) a possible final deterministic _ purification - like _ process , bringing again the particle to a unit distance from the center of the sphere @xcite .
in the previous two sections we have shown that not only the standard bloch sphere representation of two - state systems , like spin-@xmath41 entities , can be generalized to describe general @xmath4-state systems , like spin-@xmath2 entities with @xmath141 , but also that it can be extended by means of a hidden - measurement mechanism to include a full representation of all possible measurements . we have therefore answered the question regarding the nature of the interaction between the quantum and classical theaters ( an answer which constitutes a solution to the measurement problem ) . in this section ,
we want to answer the second question we have addressed in the introduction , regarding the general relation between spin eigenstates and space directions , by exploiting the descriptive power of the extended bloch model .
we have already observed that the classical - like image of a quantum spin , as a set of undetermined classical angular momentum vectors lying on a cone , is inconsistent , but we have also seen that for a spin-@xmath41 entity there is a direct one - to - one correspondence between spin eigenstates and directions of space .
these directions , however , are not anymore those characterizing classical angular momenta , but those for which the states produce a predetermined outcome , in spin measurements along those same directions .
now , even beyond the @xmath39 case , eigenstates will continue to be associated with space directions , as is clear that they are the eigenstates of spin observables that can only be defined by specifying a space direction ( the orientation of the stern - gerlach apparatus ) .
however , we can not expect anymore each eigenstate to be entirely characterized by a single space direction , as is clear that more than a single eigenvalue can be observed for each direction .
in addition to that , we know that in the extended bloch model , states are not anymore represented by three - dimensional vectors , beyond the special spin-@xmath41 situation .
so , how do spin eigenstates relate , in general , to the directions of our 3-dimensional space , that is , to the directions defining the spin observables for which they are the eigenstates ?
the answer to this question is contained in the following : + * proposition 1*. _ given a spin-@xmath2 entity , with @xmath141 , @xmath265 , and a spin observable @xmath266 , oriented along the space direction @xmath267 , then to its @xmath6 eigenstates @xmath268 , @xmath269 , represented in the extended bloch sphere by the unit vectors @xmath270 , @xmath269 , we can associate a unit vector @xmath271 , defined by : @xmath272 such that when the three - dimensional unit vector @xmath267 runs through all the possible directions of the euclidean space , @xmath273 equally spans , in an isomorphic way , all the points at the surface of a @xmath223-dimensional sub - ball of @xmath167 . _ + to prove the proposition , we write @xmath274 , where @xmath275 , @xmath276 , and the @xmath12 , @xmath13 , are the orthonormal vectors defining the canonical basis of the 3-dimensional euclidean space ( we have introduced the `` hat notation '' to distinguish 3-dimensional vectors from @xmath166-dimensional ones ) .
we define @xmath277 , and we also consider the eigenstates @xmath278 , of @xmath279 , and the associated vector @xmath280 .
we want to calculate the scalar product : @xmath281 using the spectral decompositions @xmath282 and @xmath283 , we have : @xmath284 where for the penultimate equality we have used the fact that the @xmath148 are traceless , and for the last equality that they obey @xmath151 .
so , ( [ vectorv2 ] ) becomes : @xmath285 to evaluate the right hand side of ( [ vectorv3 ] ) , we observe that the three spin operators @xmath286 , @xmath287 and @xmath72 , associated with orthogonal directions of space , are orthogonal operators : @xmath288 , if @xmath213 . to see this ,
let @xmath289 be the unitary operator representing a rotation around axis-@xmath290 , such that @xmath291 .
as we have @xmath292=0 $ ] , from the cyclicity of the trace it follows that : @xmath293 , and therefore we find that @xmath288 .
therefore : @xmath294 finally , comparing ( [ vectorv4 ] ) with ( [ vectorv3 ] ) , we obtain the equality : @xmath295 showing that the scalar product between the two vectors @xmath296 and @xmath280 , belonging to @xmath167 , is isomorphic to the scalar product between the corresponding vectors @xmath267 and @xmath297 , belonging to @xmath298 , describing the spatial orientation of the associated spin measurement apparatuses .
+ the following remarks are in order .
the vectors @xmath273 are in a one - to - one correspondence with the space directions @xmath267 .
therefore , we can say that they represent these directions within the quantum blochean theater .
we observe that : @xmath299\\ & = & d_n \left(\nu + { \nu\over n-1}\right ) = { nd_n\over n-1}\,\nu = { 1\over n-1}\sqrt{12\over n+1}\,\nu,\end{aligned}\ ] ] where for the second equality we have used @xmath300 , for @xmath301 . for @xmath39 ( @xmath146 ) , we have @xmath302 , i.e. , @xmath303 , meaning that the spin eigenstates are entirely characterized by the space directions .
on the other hand , for @xmath44 ( @xmath139 ) , the eigenvectors are not anymore oriented along the space directions @xmath273 ( we abusively say `` space directions , '' while it is clear that we should say , more precisely , `` representatives of the space directions @xmath267 within @xmath167 '' ) .
however , they all have a well - defined and fixed orientation with respect to @xmath273 , which depends on their eigenvalue and which is such that , when orthogonally projected onto @xmath273 , they give rise to equally spaced points ( see fig . [ vectorvfigure ] for the @xmath304 case ) , which are somehow reminiscent of the spots observed on the screen of a stern - gerlach apparatus , along the magnetic field direction .
this does nt mean , however , that when @xmath273 spans a three - dimensional sub - ball of @xmath167 , correspondingly to as @xmath267 spans @xmath305 , the vertex vectors @xmath306 will also span the surfaces of @xmath223-dimensional sub - balls . indeed , there are no unit sub - balls filled with states within @xmath167 ( see below ) .
the `` space '' vectors @xmath273 have a simple expression in the @xmath146 and @xmath158 cases .
indeed , for @xmath146 , we have : @xmath307 , and for @xmath158 , we have : @xmath308 ( see fig . [ vectorvfigure ] ) . in other terms , for elementary spin-@xmath309 fermions and spin-1 bosons , the space directions @xmath273 are simply given by the difference of the two eigenvectors of extremal eigenvalues . for spin-@xmath41 entities
this is so because there are only two eigenvectors available , whereas for spin-1 entities it is because one of the eigenvectors is associated with a zero eigenvalue , and therefore can not contribute to the sum ( [ vectorv ] ) .
note that the fact that @xmath273 is given by the normalized difference of two eigenvectors , means that the associated edge of the measurement simplex is oriented like @xmath273 .
equation ( [ scalarproduct ] ) immediately implies that @xmath273 is a unit vector .
one may wonder if it can be associated with a physical state .
a priori , there is no reason for this , as a ( normalized ) superposition of states within the bloch sphere will generally not be a state , beyond the @xmath146 case . and in fact , if @xmath139 , @xmath273 is never representative of a vector - state .
to see this , let @xmath310 .
if @xmath311 would be representative of a state , we would have @xmath312 , as is clear that the transition probability from a state to another state must always be a positive number .
however : @xmath313=
{ 1\over n}\left[1 + ( n-1 ) d_n\sum_{\nu =- s}^s { \bf n}_\nu\cdot{\bf n}_\mu \right ] \nonumber\\ & = & { 1\over n}\left[1 + ( n-1 ) d_n(\mu + { 1\over n-1}\,\mu ) \right]={1\over n}\left[1 + \sqrt{12\over n+1}\,\mu\right ] , \label{tracedips}\end{aligned}\ ] ] so , for the extremal value @xmath314 , we have : @xmath315 which means that @xmath316 , if @xmath139 , implying that @xmath273 can not be representative of a bona fide vector - state .
in fact , and more generally , one can prove that in the @xmath216-dimensional subspace generated by eigenvectors @xmath306 , @xmath317 , these are the only _ unit _ vectors that can be associated with physical states ( see @xcite for a proof of this statement ) .
in the present section we explore the relation between spin eigenstates and space directions in the case of a spin entity formed by two sub - entities .
we have the following : + * proposition 2*. _ given two spin entities , of spin @xmath318 and @xmath319 , and the spin observables @xmath320 and @xmath321 , oriented along the space direction @xmath267 , acting on @xmath322 and @xmath323 , respectively , and given the total spin observables @xmath324 and @xmath21 , acting on @xmath325 , then , to the @xmath326 eigenstates @xmath327 , such that @xmath328 and @xmath329 , @xmath330 , @xmath331 , represented in the extended bloch sphere by the unit vectors @xmath332 , we can associate a unit vector @xmath271 , defined by : @xmath333 such that when the three - dimensional unit vector @xmath267 runs through all the possible directions of the euclidean space , @xmath273 equally spans , in an isomorphic way , all the points at the surface of a @xmath223-dimensional sub - ball of @xmath167 . _
+ the proof follows almost identically that of _ proposition 1_. we observe that : @xmath334 and since the trace of the tensor product of two matrices is the product of their traces , we have @xmath335 , so that : @xmath336 instead of ( [ vectorv3 ] ) , we thus have : @xmath337 and instead of ( [ vectorv4 ] ) , we have : @xmath338 comparing ( [ vectorv4-bis ] ) with ( [ vectorv3-bis ] ) , we then obtain the isomomorphism ( [ scalarproduct ] ) , that is : @xmath339 .
+ the following remarks are in order . in the previous section we have observed that , for non - composite spin entities , the @xmath39 and @xmath304 cases were special , in the sense that the difference of the two extremal eigenstates ( corresponding to one of the edges of the measurement simplex ) is oriented like @xmath273 .
when we combine two spin-@xmath41 entities ( @xmath340 ) , we have four eigenstates , but two of them have zero eigenvalue : the singlet state @xmath341 and the triplet state @xmath342 , which therefore can not contribute to the sum ( [ vectorv2-bis ] ) .
this means that also in this case @xmath273 is simply given by the difference of the two extremal spin states : @xmath343 .
we also observe that : @xmath344 , which means that the eigenvector - states @xmath345 , when orthogonally projected onto @xmath273 , give rise also in this case to equally spaced points .
however , since there are now degenerate eigenvalues , although all the vectors @xmath345 are oriented differently within @xmath167 , as they are the vertex vectors of a @xmath216-simplex , those associated with the same eigenvalue will produce the same spot , when projected onto the @xmath273-axis ( see fig .
[ correspondence2 ] ) .
in the introductory section , we have explained that because of our human condition within reality , we can only explore its content from the perspective of our euclidean theater .
therefore , our situation is in a sense similar to that described by the philosopher _
plato _ in his famous allegory of the cave . however , plato opposed the material world known to us through our senses ( described as the shadows on the cave s wall ) to a world of immutable `` ideas , '' possessing an ultimate reality .
different from plato , the view we bring forward in this article is that although our euclidean theater is like a plato s cave , being the expression of a limited perspective , different caves possibly exist , corresponding to the different vantage points on reality that in principle can be adopted .
what we have called the `` quantum theater '' is in fact nothing but another cave , one not directly inhabited by us humans , and we could generally say that the whole of reality is a construction about the different possible caves ( or theaters , we use these two terms here as synonyms ) .
our work , as investigators and participators of reality , is then not only that of identifying the content of the cave we inhabit , and of the other possible caves , but also the relations ( the partial morphisms ) that exist between their different elements of reality .
this means that we do nt necessarily ask reality to be fully contained in a single fundamental theater , although of course we can not exclude that such an ultimate stage could nt one day be identified .
we neither can not exclude that there may be elements of reality belonging to one cave that can not find a direct correspondence in another cave .
we already know that certain caves are , in a sense , bigger than others .
for instance , most physicists believe that the quantum cave should contain the classical one , but this can not be taken for granted , as we know that there are severe structural shortcomings in the quantum representation . for example , in the standard quantum formalism , entities that are fully separated in experimental terms can not be described , whereas there are plenty of separated entities in the classical theater @xcite . also , there are elements of reality which appear to be of a genuine intermediate nature , which can not be represented neither in the classical nor in the quantum caves , as they belong to a truly intermediate representation , which so far has not received much attention @xcite . and
it may very well be that the many failed tentative to unify gravitational and quantum elements of reality , within a same consistent big theater , could be due to the fact that , for structural reasons , a single `` quantum & gravitational '' theater can not be constructed .
considering our description of the measurement process , we should also ponder that there is an additional difficulty to take into consideration , when different theaters , or caves , are put in relation with each other .
plato , in his allegory , used the concept of `` shadows , '' that the entities outside of our human cave casted onto its lower dimensional walls .
the observation of shadows , however , implicitly assumes a classical modality of observation , only involving a discovery aspect .
this is the modality of observation that physicists know to apply , at least in principle , when classical entities are observed by means of other classical entities . but as we have seen in sec .
[ the extended bloch sphere ] and sec .
[ general measurement ] , when quantum entities are observed , a creation aspect is also involved , i.e. , the observation can also partly produce the observed property .
this last remark is crucial in order to understand how the correspondences between elements of reality belonging to different caves should be understood . in that respect ,
let us make precise that what we mean here by the notion of `` element of reality '' is exactly what was meant by _ einstein _
, _ podolsky _ and _ rosen _ , in their famous 1935 article @xcite .
an element of reality is a _ state of prediction _ : a property of an entity that we know is _ actual _ , in the sense that , should we decide to observe it ( i.e. , to test its actuality ) , the outcome of the observation would be certainly successful .
so , what we have to consider is that the way we observe a classical entity is different form the way we observe a quantum one . more specifically , and coming back now to the example of spins , which more particularly concerns us in this article ,
when we compare a quantum spin with a classical spin , i.e. , with an angular momentum understood as a `` state of rotation in the three - dimensional euclidean space , '' we need not to forget that their elements of reality are operationally defined in a different way . indeed , for a quantum spin an observation can be invasive , unless the entity is precisely in an eigenstate of the observation in question , whereas for a classical spin the observation is always non - invasive , independently of the state of the entity . in other terms ,
when we look for those elements of reality belonging to the quantum cave that can be put in a correspondence with the elements of reality belonging to the classical one , only eigenstates can be considered , and more precisely eigenstates in association with their specific measurements .
this because only for eigenstates the observational process is of the pure discovery kind , as is the case for all classical elements of reality , and of course it would make no sense to try to compare quantum _ potential _ properties , described by _ superpositions _ of eigenstates , with classical _ actual _ properties , as superpositions of eigenstates truly describe elements of the quantum cave that have no direct counterpart with what is contained in the classical one .
this is how we have to understand the correspondence between classical spins ( angular momenta ) and quantum spin eigenstates , as expressed in our _ proposition 1_. consider fig .
[ spinfamily ] , and more particularly the second structure from the left , representing the three eigenstates of a spin-1 entity .
each one of the three cones represents a collection of classical angular momenta of length @xmath346 ( compatibly with @xmath347 ) , whose spatial orientations are consistent with the elements of reality of the corresponding quantum spin eigenstates ( since when projected onto the central axis they give rise to vectors of length @xmath0 , 0 and @xmath348 , respectively ) , forming an equilateral triangle in the 8-dimensional bloch sphere .
the central axis of the cone corresponds to the direction of observation @xmath267 , which in the case of a quantum measurement is the orientation of the stern - gerlach magnetic field .
as we have shown , this direction of space can also be identified within the extended bloch sphere , and corresponds to the ( here 8-dimensional ) vector @xmath273 ( see fig .
[ correspondence ] ) .
this means that when we move the cones in the euclidean space , the simplex of eigenstates will move accordingly , in the generalized blochean space of directions , in order to preserve the correspondence between the different elements of reality .
the correspondence , however , remains meaningful only if a single measurement is considered at a time , as is clear that eigenstates give rise to predictable outcomes only with respect to a given measurement ( and to those other measurements that are compatible with it ) .
when considering composite entities , the situation remains essentially the same , as we can still identify a vector @xmath273 which is the referent , in the quantum theater , of the spatial direction @xmath267 , as we have shown in _ proposition 2_. in this case , however , the eigenstates of the composite entity can not always be understood as resulting from combinations of the classical elements of reality associated with the composing entities . considering the simple case of two spin-@xmath41 entities ( see fig .
[ correspondence2 ] ) , we observe that the two bloch vectors @xmath349 , associated with the _ product _ ( non - entangled ) states @xmath350 , can always be associated with classical angular momentum vectors of length @xmath351 ( compatibly with @xmath352 ) , such that their projections along @xmath267 produce the values @xmath47 ( compatibly with @xmath353 ) , and their resultant is a vector of length @xmath346 ( compatibly with @xmath354 ) whose projection along @xmath267 gives @xmath355 ( compatibly with @xmath356 ) ; see fig .
[ composingvectors ] .
on the other hand , the correspondence is not anymore complete for the bloch vectors @xmath342 and @xmath341 , which are representative of the _ entangled _ ( non - product ) states @xmath357 and @xmath358 , as is clear that they are not anymore eigenvectors of the one - entity spin observables @xmath359 and @xmath360 .
it is worth emphasizing , however , that the lack of a full correspondence between the quantum and classical theaters can also manifests when the composite system is in a state which is a product of eigenstates .
for example , the two states @xmath361 , although they are eigenstates of the one - entity operators @xmath362 , @xmath363 , @xmath364 , @xmath365 , and of the total spin operator @xmath266 , they are not however eigenstates of @xmath21 , and therefore can not be represented as classical three - dimensional vectors , as this would imply a well defined value for @xmath21 . this difficulty , of obtaining a fully consistent correspondence for all spin eigenstates , is related to the well - known fact that we can either choose @xmath366 , as a _
complete set of commuting observables _ , or @xmath367 . in the first case
, we have the basis of spin eigenvectors @xmath368 , and we can always exhibit a correspondence for the total spin , but not always a correspondence for its components . in the second case , we have the basis of spin eigenvectors @xmath369 , and we can always exhibit a correspondence for the composing spins , but not always a correspondence for the total spin .
note that the above two eigenbases share the first two of their vectors , which are precisely those eigenstates that admit a full representation , within the spatial theater , for both the total spin and for its components .
according to the above , it is clear that an alternative , equivalent version of _ proposition 2 _ can be written , using a different set of spin eigenvectors to defne the space vectors @xmath273 .
more precisely , we have : + * proposition 2-bis*. _ given two spin entities , of spin @xmath318 and @xmath319 , and the spin observables @xmath320 and @xmath321 , oriented along the space direction @xmath267 , acting on @xmath322 and @xmath323 , respectively , and given the total spin observable @xmath324 , acting on @xmath325 , then , to the @xmath326 eigenstates @xmath370 , such that @xmath371 , @xmath372 , and @xmath373 , @xmath374 , @xmath375 , represented in the extended bloch sphere by the unit vectors @xmath376 , we can associate a unit vector @xmath377 , defined by : @xmath378 such that when the three - dimensional unit vector @xmath267 runs through all the possible directions of the euclidean space , @xmath379 equally spans , in an isomorphic way , all the points at the surface of a @xmath223-dimensional sub - ball of @xmath167 . _ + the proof of _ proposition 2-bis _ is identical _ mutatis mutandis _ to that of _ poroposition 2_. one may wonder if the vectors @xmath273 defined in ( [ vectorv2-bis ] ) are the same as the vectors @xmath379 defined in ( [ vectorv2-tris ] ) , considering that they both obey ( [ scalarproduct ] ) .
the answer is affirmative , and to show this it is sufficient to observe that : @xmath380 where for the second equality we have used that @xmath381 . in other terms , modulo a renormalization factor , the components of @xmath379 provide the expansion of @xmath266 , on the basis of the generators of @xmath140 .
but since we also have : @xmath382 and the expansion of @xmath266 is unique , we can conclude that the vectors ( [ vectorv2-bis ] ) and ( [ vectorv2-tris ] ) are exactly the same .
in this article we have presented the extended bloch representation of quantum mechanics and the associated hidden - measurement interpretation , and used it to explore spin states and spin measurements .
the extended bloch representation is a candidate for a refined version of quantum theory , in which operator - states ( density matrices ) can also play a role as pure states in the description of what goes on during a measurement process .
we refer the reader to @xcite for a more detailed and complete description of the model , and for a further generalization of it .
let us mention that it is also possible to relax the hypothesis that the substance filling the measurement simplexes is uniform ( in the sense of having a uniform probability density of disintegrating in a point ) , and still be able to derive the born rule , if a ( universal ) average over all possible non - uniform substances is considered .
also , one can show that non - uniform substances , when not averaged out , describe physical entities of a genuine intermediate nature , which can not be described by the classical or quantum formalisms ( their elements of reality being neither contained in the classical nor in the quantum theaters ) @xcite .
the extended bloch representation elucidates the mechanism through which quantum entities interact with classical measurement apparatus .
the interaction is invasive and governed by ( non - spatial ) fluctuations , which by definition can not be controlled by the experimenter , and therefore produce a genuine indeterministic change of the state of the measured entity .
according to this interpretation , a quantum measurement is not only a process of discovery , but also , and above all , a process of creation .
the extended bloch representation also allows us to gain some insight into how the vaster quantum world relates to our ordinary `` niche '' reality , made of macroscopic objects which are always present in the three - dimensional euclidean theater . in that respect
, we have shown that quantum spins can not be considered as intrinsic angular momenta , as no space directions can be directly associated to them , beyond the @xmath39 situation .
on the other hand , they always remain in a specific relation to the space directions , as expressed by the fact that they possess a predetermined orientation with respect to the vectors ( [ vectorv ] ) and ( [ vectorv2-bis])-([vectorv2-tris ] ) , which are the representative of the space directions within the blochean theater .
it is worth observing that even though spin-@xmath41 entities can be characterized , when isolated , by single space directions , not for this they should be considered to be generally in space , as is clear that @xmath144 is a _
double cover _ of @xmath145 , which means that @xmath383 and @xmath384 rotations , although they correspond to two distinct elements of @xmath144 , are mapped onto the same element of @xmath145 . in the bloch sphere
this two - to - one correspondence can not be seen , as global phase factors are not represented into it .
however , when a spin is coupled with the translational degrees of freedom of a microscopic entity , the distinction between @xmath383 and @xmath384 rotations can become observable , in the suitable experimental context .
this is precisely what was achieved by helmut rauch and his group , in their celebrated experiments of interferometry @xcite , where neutrons , passing one by one through a perfect silicon crystal ( made of three parallel lips of same thickness ; see fig .
[ rauch - figure ] ) , were allowed to interfere with themselves , because of the spatial splitting and subsequent recombination of their wave functions .
this was done in a way so as to allow the experimenter to only act on one of the two spatially separated components , by means of a magnetic field whose intensity could be varied , so producing an interference contribution in the detected intensity , with a remarkable @xmath384-periodic behavior ( when expressed as a function of the rotation angle of the neutron s spin ) .
rauch s experiment has shown us two things : the first one is that , as we said , even spin-@xmath41 entities are not in space , considering that a @xmath383 rotation is not sufficient to bring them back exactly in the same state , when not in an isolated condition .
in fact , this was already clear from our discussion in the introduction , considering that we could not find classical elements of reality able to account for spin observations along all possible directions .
and this also emerged in our analysis of the combination of two spin@xmath41 entities , which although they can be in a well defined eigenstate within the composite entity ( when in a product state ) , not for this they can always be associated with a well - defined three - dimensional classical angular momentum vectors .
of course , if , strictly speaking , spins are not in space , it means that it is not totally correct to interpret the action of the magnetic field on the spinor component of the wave function in rauch s experiments as a true rotation in space . and
in fact , it has been proposed ( also by one of us ) that a minimal interpretation of the experiment is as a longitudinal stern - gerlach effect , i.e. , that the ( apparent ) larmor precession would only result from the interference between the two stern - gerlach states in the weak field limit @xcite .
rauch s experiment has also shown us that the notion of non - spatiality equally applies to the configuration space part of the wave function .
indeed , it shows that although a neutron can only be detected in one place , it nevertheless can be acted upon , simultaneously , from different separated places , in a way that can produce a measurable effect . and
if we reflect attentively , such a possibility can only be understood if we admit that a neutron is an entity which is not permanently present in space , but only possibly enters it when it interacts with a measuring apparatus , which in a sense is able to `` drag '' it into a spatial state @xcite . from a conceptual point of view
, we should certainly distinguish the notion of _ space direction _ from the notion of _ space location_. elementary entities are generally to be considered not in space in both senses : because they do nt generally have an actual position in space , but only a potential one , and because their spin ( if they are not zero ) are not oriented toward a space direction , apart the special case of isolated spin-@xmath41 entities ( where the term `` isolated '' means here not coupled with other systems , including the translational degrees of freedom of the very entity carrying the spin ) .
now , as we said , microscopic entities are able to acquire a location in space , in specific contexts , as is clear that we can localize them in space with arbitrary precision . on the other hand ,
their spins , apart the @xmath39 case , remain always out of alignment with respect to space , as is clear that all spin eigenvectors within the bloch sphere are oriented in directions which are different from those of the vectors @xmath273 , defined in _ proposition 1 , 2 _ and _ 2-bis _ , which are the referents of the space directions within the quantum theater . in the special @xmath39 situation ,
the space directions and the directions of the spin eigenstates coincide , but as we have seen the correspondence between spin eigenvectors and angular momentum vectors is incomplete : because it can only be established for a single spin measurement at a time , and because it does nt fully work when composite entities are considered .
therefore , even spin-@xmath41 should not be considered to be in space ( or fully in space ) , i.e. , to be representable by one ( or many ) three - dimensional vectors of suitable length and orientation .
but although generally not in space , quantum spin entities , when in an eigenstate , are nevertheless always in a specific relation to space , as evidenced by the fact that the eigenvectors in the bloch sphere always maintain fixed and predetermined orientations with respect to the `` spatial '' vectors @xmath273 . these orientations are precisely those that allow the eigenvectors to always be positioned at the vertices of the measurement simplexes , so that they can not be affected by the collapse of the corresponding membranes .
this is why we can predict in advance the outcome of the measurements , and bring them into correspondence with the classical angular momentum elements of reality of the classical theater .
it is in that sense , and only in that sense , that we can justify the assertion that , say , an electron has a spin of a given magnitude along a given direction .
however , as emphasized many times , when we combine spins , the classical picture breaks down , as is clear that there is no simple relation between the elements of reality described by two 3-dimensional bloch spheres , associated with two separated spin-@xmath41 entities , and the 15-dimensional bloch sphere describing the states emerging from their quantum combinations .
this because no sum of three - dimensional vectors will ever be able to account for the emergence of a 15-dimensional bloch sphere from two 3-dimensional ones , and of a tetrahedron ( 3-simplex ) from two line segments ( 1-simplexes ) .
a final comment is in order . as we explained in sec .
[ connecting elements of reality ] , not all the elements of reality that are present in the quantum theater do necessarily have a correspondence , however partial , with some of the elements of reality present in the classical theater .
quantum spins can individually be observed in the classical theater also because they are all in a specific relation to space directions , i.e. , to classical elements of reality . but this needs not to be the case for other quantum properties , like for instance the color charge properties of individual quarks , which , as far as we know , and contrary to spins , have no relations to space directions , which may be one of the reasons of their confinement within the quantum cave .
this is in accordance with the view of realism that we have adopted in this article , that we will call _ multiplex realism_. it is worth emphasizing that multiplex realism is not instrumentalism in disguise , nor a new form of model - dependent realism @xcite , or of structural realism @xcite .
also , the view should not be considered as the expression of the belief that we necessarily live in a multiplex reality .
this is certainly a possibility , but it should be clear that the main reason for adopting such a concept of realism is that we give importance to the following three observations .
\(1 ) due to our presence , for hundreds of thousands of years , in that specific niche of reality which is the surface of our planet earth , surrounded by material objects that with good approximation obey the newton s laws , we have constructed an euclidean theater to stage our relations and interactions with these classical entities .
\(2 ) in more recent times , we became aware of the existence of quantum and relativistic entities , whose reality could only be put in a partial correspondence with the properties and behaviors of the previously known euclidean entities , that is , we could only find _ partial morphisms _ , and not perfect isomorphisms .
for instance , quantum eigenstates could be linked , to some extent , to classical elements of reality , but not their superpositions .
\(3 ) although we have good reasons to believe that the quantum and relativistic theaters are the expression of more advanced theories of reality , we nonetheless continue to be strongly influenced by our euclidean theater , also because we are ordinarily not aware that it has been constructed by our ancestors , in particular as regards the primitive notions of space and time . therefore , we have interest in becoming fully aware of this ` construction aspect , ' taking into consideration its effect in the way we conceive , and try to understand , the more recently discovered non - euclidian quantum and relativistic elements of reality . however , it is important to emphasize that realism can be multiplex even if reality is singleplex . by this
we mean that multiplex realism should not to be intended as a statement about the multiple nature of our reality , but as an approach to reality which considers that certain incompatibilities , like that between relativity and quantum theories , can be more fruitfully studied if we take into account the fact that their elements of reality are also , in part , the result of a constructive process .
d. aerts , `` the unification of personal presents : a dialogue of different world views , '' pp . 6382 . in : _
international readings on theory , history and philosophy of culture : ontology of dialogue _ , vol.12 ( 2002 ) . w. gerlach and o. stern , `` das magnetische moment des silberatoms , '' zeitschrift
fr physik 9 , 353355 ( 1922 ) .
h. rauch , w. treimer , u. bonse , `` test of a single crystal neutron interferometer , '' phys . lett .
a 47 , 369 ( 1974 ) .
h. rauch et al . , verification of coherent spinor rotations of fermions , " phys .
54a , 425427 ( 1975 ) .
h. rauch , neutron interferometric tests of quantum mechanics , " helv .
acta 61 , 589 ( 1988 ) .
a. aspect et al .
, `` experimental realization of einstein - podolsky - rosen - bohm gedankenexperiment : a new violation of bell s inequalities , '' phys .
lett . , 49 , p. 91
a. aspect , `` bell s inequality test : more ideal than ever , '' nature ( london ) , 398 , p. 189
k. hentschel , `` spin , '' pp . 726731 . in : _ compendium of quantum physics _
, d. greenberger , k. hentschel and f. weinert ( eds . ) , springer - verlag , berlin heidelberg ( 2009 ) . h. poincar , _ thorie mathematique de la lumire _ , gauthiers - villars , paris , vol . 2 ( 1892 ) .
f. bloch , `` nuclear induction , '' phys .
70 , 460474 ( 1946 ) .
d. aerts , `` a possible explanation for the probabilities of quantum mechanics , '' journal of mathematical physics 27 , 202210 ( 1986 ) .
d. aerts , `` the origin of the non - classical character of the quantum probability model . '' in : information , complexity , and control in quantum physics , eds .
a. blanquiere et al , springer - verlag , berlin ( 1987 ) .
d. aerts and m. sassoli de bianchi , `` the extended bloch representation of quantum mechanics and the hidden - measurement solution to the measurement problem , '' annals of physics 351 , 9751025 ( 2014 ) .
doi : 10.1016/j.aop.2014.09.020 .
d. aerts , `` the entity and modern physics : the creation - discovery view of reality . '' in : _ interpreting bodies : classical and quantum objects in modern physics _ , ed .
castellani , e. princeton unversity press , princeton ( 1998 ) .
d. aerts , `` the stuff the world is made of : physics and reality , '' pp
. 129183 . in : _
the white book of ` einstein meets magritte ' _ , edited by diederik aerts , jan broekaert and ernest mathijs , kluwer academic publishers , dordrecht , 274 pp .
a. m. gleason , `` measures on the closed subspaces of a hilbert space , '' j. math . mech . 6 , 885893 ( 1957 ) .
s. kochen and e. p. specker , `` the problem of hidden variables in quantum mechanics , '' j. math .
17 , 5987 ( 1967 ) .
d. aerts , b. coecke , b. d. hooghe and f. valckenborgh , `` a mechanistic macroscopic physical entity with a three - dimensional hilbert space description , '' helv .
acta 70 , 793 ( 1997 ) .
b. coecke , `` hidden measurement representation for quantum entities described by finite dimensional complex hilbert spaces , '' found .
phys . , 25 , 1185 ( 1995 ) . b. coecke , `` generalization of the proof on the existence of hidden measurements to experiments with an infinite set of outcomes , '' found .
8 , 437 ( 1995 ) .
arvind , k. s. mallesh and n. mukunda , `` a generalized pancharatnam geometric phase formula for three - level quantum systems , '' j. phys .
a 30 , 2417 ( 1997 ) g. kimura``the bloch vector for @xmath4-level systems , '' phys .
lett . a 314 , 339 ( 2003 ) . m. s. byrd and n. khaneja , `` characterization of the positivity of the density matrix in terms of the coherence vector representation , '' phys .
a 68 , 062322 ( 2003 ) .
g. kimura & a. kossakowski , `` the bloch - vector space for n - level systems the spherical - coordinate point of view , '' open sys .
information dyn .
12 , 207 ( 2005 ) .
i. bengtsson and k. yczkowski , _ geometry of quantum states : an introduction to quantum entanglement _ , cambridge university press , cambridge ( 2006 ) . i. bengtsson and k. yczkowski , `` geometry of the set of mixed quantum states : an apophatic approach , '' pp . 175197 . in : _ geometric methods in physics ,
xxx workshop 2011 , trends in mathematics _ , springer ( 2013 ) .
l. p. hughston , r. jozsa and william k. wootters,``a complete classification of quantum ensembles having a given density matrix , '' physics letters a 183 , 1418 ( 1993 ) .
f. t. hioe , j. h. eberly , `` @xmath4-level coherence vector and higher conservation laws in quantum optics and quantum mechanics , '' phys .
47 , 838841 ( 1981 ) .
r. alicki , k. lendi , _ quantum dynamical semigroups and application _ , lecture notes in physics vol .
286 , springer - verlag , berlin ( 1987 ) .
g. mahler , v. a. weberruss , _ quantum networks _ , springer , berlin ( 1995 ) .
d. aerts `` description of many physical entities without the paradoxes encountered in quantum mechanics , '' found .
12 , 11311170 ( 1982 ) .
d. aerts , `` the missing elements of reality in the description of quantum mechanics of the epr paradox situation , '' helvetica physica acta , 57 , 421428 ( 1984 ) .
d. aerts , d. `` foundations of quantum physics : a general realistic and operational approach , '' international journal of theoretical physics 38 , 289358 ( 1999 ) .
d. aerts , `` the description of joint quantum entities and the formulation of a paradox , '' international journal of theoretical physics 39 , 485496 ( 2000 ) .
a. einstein , b. podolsky and n. rosen , `` can quantum - mechanical description of physical reality be considered complete ?
, '' phys .
47 , 777 ( 1935 ) .
d. aerts and m. sassoli de bianchi , `` the unreasonable success of quantum probability i : quantum measurements as uniform measurements , '' arxiv:1401.2647 [ quant - ph ] ( 2014 ) .
d. aerts and m. sassoli de bianchi , `` the unreasonable success of quantum probability ii : quantum measurements as universal measurements , '' arxiv:1401.2650 [ quant - ph ] ( 2014 ) .
d. aerts , `` an attempt to imagine parts of the reality of the micro - world , '' 325 , in _ problems in quantum physics ii _ ; gdansk 89 , edited by : j. mizerski et al . ,
world scientific publishing company , singapore ( 1990 ) .
d. aerts , `` a mechanistic classical laboratory situation violating the bell inequalities with @xmath385 , exactly ` in the same way ' as its violations by the epr experiments , '' helv .
123 ( 1991 ) . m. sassoli de bianchi , `` using simple elastic bands to explain quantum mechanics : a conceptual review of two of aerts machine - models , '' centr .
11 , 147161 ( 2013 ) .
f. mezei , `` zeeman energy , interference and neutron spin echo : a minimal theory , '' physica b 151 , 7481 ( 1988 ) .
a. martin and m. sassoli de bianchi , `` spin precession revisited , '' found .
24 , 13711378 ( 1994 ) .
s. hawking and l. mlodinow , _ the grand design _ , bantam books , new york ( 2010 ) .
j. worrall , j. , 1989 .
`` structural realism : the best of both worlds ? '' dialectica 43 , 99124 . reprinted in d. papineau ( ed . ) , _ the philosophy of science _ , oxford university press , oxford , pp .
139165 ( 1989 ) . | the standard bloch sphere representation has been recently generalized to describe not only systems of arbitrary dimension , but also their measurements , in what has been called the _ extended bloch representation of quantum mechanics_. this model , which offers a solution to the longstanding measurement problem , is based on the _ hidden - measurement interpretation of quantum mechanics _ , according to which the born rule results from our lack of knowledge of the measurement interaction that each time is actualized between the measuring apparatus and the measured entity . in this article we present the extended bloch model and
use it to investigate , more specifically , the nature of the quantum spin entities and of their relation to our three - dimensional euclidean theater .
our analysis shows that spin eigenstates can not generally be associated with directions in the euclidean space , but only with generalized directions in the blochean space , which apart from the special case of spin one - half entities , is a space of higher dimensionality . accordingly
, spin entities have to be considered as genuine non - spatial entities .
we also show , however , that specific vectors can be identified in the blochean theater that are isomorphic to the euclidean space directions , and therefore representative of them , and that spin eigenstates always have a predetermined orientation with respect to them .
we use the details of our results to put forward a new view of realism , that we call _ multiplex realism _ , providing a specific framework with which to interpret the human observations and understanding of the component parts of the world .
elements of reality can be represented in different theaters , one being our customary euclidean space , and another one the quantum realm , revealed to us through our sophisticated experiments , whose elements of reality , in the quantum jargon , are the eigenvalues and eigenstates .
our understanding of the component parts of the world can then be guided by looking for the possible connections , in the form of partial morphisms , between the different representations , which is precisely what we do in this article with regard to spin entities .
: spin eigenstate , spin measurement , extended bloch sphere , hidden - measurement interpretation , non - spatiality , multiplex realism |
tde is an astronomical phenomenon that occurs when a star gets too close to a supermassive black hole in the galaxy center and is disrupted by the tidal force of the black hole .
part of stellar material is bound and accreted by the central black hole , resulting in a bright optical , uv and soft x - ray emission ( * ? ? ?
* ; * ? ? ?
* and references therein ) .
there are growing number of candidate tdes being discovered in soft x - ray , ultraviolet and optical surveys , see @xcite for a recent review .
recently , three unusual tde candidates are discovered by _ swift _ , i.e. swift j164449.3 + 573451 swift j2058.4 + 0516 swift j1112.2 - 8238 ( hereafter sw j1644 + 57 , sw j2058 + 05 and sw j1112 - 82 for short respectively ) , which have very bright non - thermal hard x - ray and radio emissions @xcite .
the luminous non - thermal x - ray and radio emissions are thought to be produced by relativistic jets @xcite .
sw j1644 + 57 shows a highly variable light curve in x - rays , as observed by x - ray telescope ( xrt ) onboard _
swift_. at redshift @xmath6 , the isotropic luminosity of the x - ray emission is as high as @xmath7 erg / s .
sw j2058 + 05 exhibits a luminous , long - lived x - ray outburst with an isotropic peak luminosity of @xmath8 ( at redshift @xmath9 ) .
its total isotropic energy ( 0.3 - 10 kev ) on timescale of the first two months amounts to @xmath10 erg .
sw j1112 - 82 was initially also discovered by swift / bat ( burst alert telescope ) in june 2011 as an unknown , long - lived ( order of days ) @xmath11-ray transient source .
it exhibits a similar bright x - ray flare and its position is consistent with the nucleus of a faint galaxy at @xmath12 @xcite .
the peak x/@xmath11-ray luminosity of sw j1112 - 82 exceeds @xmath13 .
the non - thermal x - ray emission is thought to be produced by synchrotron radiation of relativistic electrons .
one would naturally expect inverse - compton scattering emission from the same electrons , which may produce high - energy gamma - ray emission .
as the three tdes with relativistic jets emit a total isotropic energy of about @xmath10 erg in x - ray band , comparable to or even larger than that in the grb prompt emission and blazars flares , one would expect the high - energy gamma - ray emission detectable by _ fermi_/lat .
motivated by this , we search for high - energy gamma - ray emission from these three relativistic tdes .
asassn-14li is a normal optically - discovered tde at distance of about 90 mpc @xcite .
transient radio emission has been detected from this event and modeling of the radio emission gives a kinetic energy of @xmath14 erg in a non - relativistic or mildly relativistic outflow @xcite .
the relativistic electrons producing radio emission may in principle also produce high - energy gamma - ray emission , although the flux level depends on the energy of these electrons and the strength of the magnetic field .
it was also proposed that the unbound debris after the disruption will encounter the interstellar medium or dense molecular clouds around the central massive black hole , producing high - energy @xmath15-ray afterglow through @xmath16 collisions .
thus , we also search for high - energy gamma - ray emission from the normal tde asassn-14li .
the analysis of the _ fermi _ lat data of tdes is presented in section 2 .
we present the result for each tde in section 3 .
furthermore , we compare the high - energy gamma - ray emission of relativistic tdes with that of grb prompt emission and blazar flares
. then in section 4 , we derive constraints on the bulk motion lorentz factor of the source based on the non - detection of the high energy emission .
finally we give a short summary . throughout this paper , we take the standard @xmath17cdm cosmology with parameters @xmath18 km s@xmath3 mpc@xmath3 , @xmath19 , and @xmath20 .
the _ fermi _ lat is a pair conversion telescope designed to cover the energy band from 20 mev to greater than 300 gev , and operates primarily in an all - sky scanning survey mode @xcite .
the lat data currently being released by the fssc ( _ fermi _ science support center ) have been processed using the `` pass 8 '' ( p8r2 ) event - level analysis , and specifically p8r2 source event class data are selected here .
the analysis is based on the lat science tool version v10r0p5 . in order to reduce the contamination from earth limb emission ,
less than @xmath21 of lat rocking angle and less than @xmath22 of local zenith angle are required .
moreover , we exclude the time period when the spacecraft is above the south atlantic anomaly .
we perform an unbinned maximum likelihood analysis of the selected data with the following method .
front - back converting photons ( evtype =3 ) of energies between 100 mev and 10 gev are divided into twenty logarithmically spaced energy bins for fitting of a power law spectrum ( @xmath23 ) .
all events in a region of interest ( roi ) of @xmath24 have been used , and point sources and extending sources within extra @xmath25 in the lat 4-year point source catalog ( 3fgl)@xcite are added to the model file .
the galactic diffuse and isotropic emission are molded with gll_iem_v06.fits and iso_p8r2_source_v6_v06.txt respectively .
when it leaves too many free parameters to gain a good spectral fit for a faint source at a short time interval , we fix the spectral form and free the normalization ( prefactor ) of faint or distant background sources .
the above data analysis thread would be made products for selected tdes .
binned analysis is considered to check the results . for the purpose of comparison
, we also analyze the _
fermi_/lat data of two typical blazars , i.e. mrk 421 and 3c 279 , during some flare episodes @xcite .
analysis threads are similar to the above method .
for grbs , we select the first _ fermi _ lat grb catalog @xcite as our sample for comparison with tdes .
the processing procedure for the lat data of grbs follows our previous work @xcite .
thanks to _ swift_/bat @xcite , for bright x - ray sources such as the three relativistic tdes mentioned above , we can get the daily average count rate in the survey mode ( data quality flag = 0 ) @xcite ( because sw j1644 + 57 was mistaken for a grb , it was observed by bat burst mode and had a good follow - up observation @xcite ) . considering the limited bat energy band ( 15 - 50 kev ) , the corresponding flux will be displayed in the sed plot ( figure [ sed3 ] ) as one data point .
the likelihood analysis of sw j1644 + 57 centered at the position ( ra . , dec.)=(@xmath26 ) results in a low ts value ( @xmath27 ) , i.e. no significant high - energy emission is found from the bat trigger time to one hundred days after the trigger ( @xmath28s ) .
its @xmath4 confidence upper limit fluxes are presented in table [ lc3bins ] for different time intervals .
we fix the spectral photon index at @xmath29 .
taking different spectral index @xmath30 will result in a slight , but insignificant difference .
we examine the archival database day by day for three days before @xmath31 and yield non detection either .
our results are consistent with the results in @xcite , in spite of some slight difference due to new data ( pass 8) , different photon index @xmath32 and background model .
the sed in 0.3 - 150 kev during the bright period is extracted from @xcite and plotted in figure [ sed3 ] .
the upper limit fluence in 0.1 - 10 gev is @xmath33 erg @xmath34 during the first three days when the source is in the brightest phase in bat . integrating the bat emission over the significant emission period
, we get a total 15 - 50 kev fluence of @xmath35 erg @xmath34 , as shown in table [ fluencecom ] . in order to compare with grbs , whose
fluences are given in the _ fermi _ gamma - ray burst monitor ( gbm ) band ( 10 - 1000 kev ) , we adopt a power law spectrum with an index @xmath36 ( @xmath37 ) to extrapolate the 15 - 50 kev fluence to the 10 - 1000 kev range for sw j1644 + 57 , we adopt -2.15 for the estimate . ] , resulting in a 10 - 1000 kev fluence of about @xmath38 .
one can see that the x/@xmath11-ray fluence is about two orders of magnitude larger than that in 0.1 - 10 gev gamma - rays .
the standard gtlike analysis of the _
fermi_/lat data of sw j2058 + 05 centered at the position ( ra . , dec.)=(@xmath39 ) gives a non detection .
the upper limit fluxes from the bat trigger time to one hundred days after trigger ( @xmath40s ) , divided into three time intervals , are given in table [ lc3bins ] and also shown in figure [ lightcurve4 ] . using the bat and lat observational data , the sed on 23 may 2011 , which is the brightest epoch in bat , is shown in figure [ sed3 ] .
sw j2058 + 05 has a seventeen - day activity in the bat observation with a total fluence of @xmath41 erg @xmath34 when extrapolated to the energy range of 10 - 1000 kev ( see section 3.0.1 ) .
the corresponding lat upper limit fluence is @xmath42 erg @xmath34 in the same time interval .
the likelihood analysis of the _
fermi_/lat data of sw j1112 - 82 centered at ( ra .
, dec.)=(@xmath43 ) also yields a non - detection .
the upper limit fluxes from the bat trigger to one hundred days after trigger ( @xmath28s ) are given in table [ lc3bins ] and also shown in figure [ lightcurve4 ] .
we show the bat and lat data on 16 june 2011 ( the brightest epoch in bat ) in the sed plot ( figure [ sed3 ] ) .
the lat upper limit fluence during the period with bright bat emission ( from 16 june 2011 to 19 june 2011 ) is @xmath44 erg @xmath34 , while the 10 - 1000 kev fluence in the same time interval is @xmath45 erg @xmath34 .
the nearest tde discovered to data , asassn-14li , is centered at ( ra .
, dec.)=(@xmath46 ) @xcite .
we analyze the _
fermi_-lat data during the period from 22 november 2014 to 22 november 2015 and find no evidence of high - energy gamma - ray emission , with a low ts value ( @xmath47 ) .
the upper limit fluxes are presented in table [ lc3bins ] and also shown in figure [ lightcurve4 ] .
search for high energy gamma - ray emission from the same source in the archival data also yields only upper limit . at distance of 90 mpc ,
the upper limit luminosity in 0.1 - 10 gev is @xmath48 erg s@xmath3 during the period from 22 november 2014 to 17 march 2015 .
this limit luminosity is lower than the soft x - ray ( 0.1 - 3 kev ) luminosity , which is about @xmath49 @xcite .
however , it is higher than the predicted gamma - ray luminosity , @xmath50 erg s@xmath3 , by , so closer tdes detected in future are needed to test the prediction . since relativistic tdes have relativistic jets similar to grbs and blazars , we compare the gev emission in tdes with the other two relativistic sources . for grbs , we make the combined gbm - lat spectral analysis during the prompt emission period , and
the data of the gev fluence and x/@xmath11-ray fluence are presented in figure [ comparison3 ] , which are consistent with the results in the _ fermi _ lat first gamma - ray burst catalog ( see their fig .
17 ) @xcite . for blazars
, we choose mrk 421 and 3c 279 as representatives of bl lac objects and flat spectrum radio quasar ( fsrqs ) respectively .
we select the time intervals that x - ray emission flares to analyze the _ fermi _ lat data @xcite .
the modeling of 0.1 - 10 gev spectrum of mrk 421 with a simple power law function gives a photon index of @xmath51 , whereas it gives a soft photon index with @xmath52 for 3c 279 , which are consistent with the statistic characteristic of joint bat - lat spectral properties of blazars @xcite .
long - time 0.1 - 300 gev data gives a preferred spectral model logarithmic parabola for 3c279 in 3fgl , so we check our result with this spectral model and find almost no difference in the fluence . the results of their x/@xmath11-ray and gev fluences are listed in table [ fluencecom ] . )
may not be appropriate , but anyway , a few times change of x/@xmath11-ray ( 10 - 1000 kev ) fluence will not make the fluence ratio below 1% in figure [ comparison3 ] . ] in figure [ comparison3 ] , we compare the x/@xmath11-ray fluence and gev fluence of the three types of relativistic sources .
we find that all grbs have gev fluences larger than 1% of the x/@xmath11-ray fluences . among them
, a significant fraction of grbs have gev fluences larger than 10% of the x/@xmath11-ray fluence .
blazar flares have similar properties and all of them have gev fluences larger than a few per cent of the x/@xmath11-ray fluences .
in contrast , the gev fluences of all three relativistic tdes are smaller than 1% of the x/@xmath11-ray fluences .
this may indicate that the high - energy gamma - ray emission in tdes are highly absorbed , as we will discuss below .
the non - thermal x - ray emission in three relativistic tdes may be produced by relativistic electrons via synchrotron radiation @xcite .
the size of the source is estimated to be @xmath53 , where @xmath30 is the bulk lorentz factor of the source , @xmath54 is the variability timescale of x - ray emission and we have assumed that the jet aligns with the light of sight of the observer .
here we use c.g.s .
units and the denotation @xmath55 in the paper .
the observed minimum variability time of the x - ray emission in sw j1644 + 17 is about @xmath56 @xcite . assuming that the magnetic field energy density is in equipartition with the radiation energy density , the magnetic field in the comoving frame of the source is estimated to be @xmath57 , where @xmath58 is the equipartition factor .
for such a magnetic field , the x - ray photons with frequency @xmath59 are produced by relativistic electrons with lorentz factors of @xmath60 .
for these electrons , we expect an inverse - compton component peaking at @xmath61 when other parameters are taken with typical values
. the ic flux should be comparable to or larger than that of the synchrotron component if @xmath62 , thus we would expect a gev component with a luminosity of @xmath63 erg s@xmath3 .
the non - detection of such a high - energy component can be attributed to a high absorption opacity due to low - energy photons in the source , i.e. the @xmath64 absorption optical depth should be @xmath65 .
the optical depth for @xmath64 absorption is given by @xmath66 , where @xmath67 is the cross section for @xmath64 absorption ( @xmath68 is the thompson cross section ) and @xmath69 is the comoving - frame number density of the target photons that interact with high - energy photons . for high energy photons with energy @xmath70 , the energy of the target photons is @xmath71 . for a power - law spectrum with a photon index of @xmath72 @xcite , the number density of target photons
is @xmath73 , where @xmath74 .
one can derive an upper limit on the bulk lorentz factor from the condition @xmath75 , i.e. @xmath76 this shows that the bulk lorentz factors in tdes are much lower than that of grbs @xcite , and may even be lower than that of some blazars .
we searched for the high - energy gamma - ray emission from tdes with the _ fermi _ lat survey data , including three tdes confirmed with relativistic jets ( sw j1644 + 57 , sw j2058 + 05 , sw
j1112 - 82 ) and the nearest normal tde asassn-14li .
no significant emission is found from these tdes during the period from the _ swift_/bat trigger time ( or the inferred disruption time ) to about one hundred days later .
compared with the bright non - thermal x - ray emission , the non - detection of high - energy emission in three relativistic tdes implies that high energy emission is seriously suppressed , most likely due to the @xmath77 attenuation by soft photons in the source .
then , we derive an upper limit on the bulk lorentz factor , @xmath1 , for relativistic jets in tdes .
the non - detection of high - energy emission from the normal tde asassn-14li gives an upper limit of @xmath78 during the first @xmath5 s after the disruption , which is already lower than the x - ray luminosity @xcite , but still higher than the predicted gamma - ray luminosity .
this work made use of data supplied by the uk swift science data center at the university of leicester , and swift / bat transient monitor results provided by the swift / bat team .
we thank jin zhang for helpful discussion .
this work is supported by the 973 program under grant 2014cb845800 , the nsfc under grants 11273016 and 11033002 , and the excellent youth foundation of jiangsu province ( bk2012011 ) .
abdo , a. a. , ackermann , m. , ajello , m. , et al .
2009 , astroparticle physics , 32 , 193 acero , f. , ackermann , m. , ajello , m. , et al .
2015 , , 218 , 23 ackermann , m. , ajello , m. , asano , k. , et al .
2013 , , 209 , 11 alexander , k. d. , berger , e. , guillochon , j. , zauderer , b. a. , & williams , p. k. g. 2015 , arxiv:1510.01226 atwood , w. b. , abdo , a. a. , ackermann , m. , et al .
2009 , , 697 , 1071 barthelmy , s. d. , barbier , l. m. , cummings , j. r. , et al .
2005 , , 120 , 143 bartoli , b. , bernardini , p. , bi , x. j. , et al
. 2015 , arxiv:1511.06851 bloom , j. s. , giannios , d. , metzger , b. d. , et al .
2011 , science , 333 , 203 brown , g. c. , levan , a. j. , stanway , e. r. , et al .
2015 , , 452 , 4297 burrows , d. n. , kennea , j. a. , ghisellini , g. , et al .
2011 , , 476 , 421 cao , d. , & wang , x .- y .
2012 , , 761 , 111 cenko , s. b. , krimm , h. a. , horesh , a. , et al .
2012 , , 753 , 77 chen , x. , gmez - vargas , g. , & guillochon , j. 2015 , arxiv:1512.06124 cheng , k. s. , chernyshov , d. o. , & dogiel , v. a. 2007 , , 473 , 351 cheng , k. s. , chernyshov , d. o. , & dogiel , v. a. 2006 , , 645 , 1138 gehrels , n. , chincarini , g. , giommi , p. , et al .
2004 , , 611 , 1005 hayashida , m. , madejski , g. m. , nalewajko , k. , et al .
2012 , , 754 , 114 holoien , t. w .- s . ,
kochanek , c. s. , prieto , j. l. , et al .
2016 , , 455 , 2918 jose , j. , guo , z. , long , f. , et al .
2014 , the astronomer s telegram , 6777 , komossa , s. 2015 , journal of high energy astrophysics , 7 , 148 krimm , h. a. , holland , s. t. , corbet , r. h. d. , et al .
2013 , , 209 , 14 krimm , h. a. , kennea , j. a. , holland , s. t. , et al .
2011 , the astronomer s telegram , 3463 , krimm , h. a. , kennea , j. a. , holland , s. t. , et al .
2011 , the astronomer s telegram , 3384 , lithwick , y. , & sari , r. 2001 , , 555 , 540 liu , d. , peer , a. , & loeb , a. 2015 , , 798 , 13 lodato , g. , franchini , a. , bonnerot , c. , & rossi , e. m. 2015 , journal of high energy astrophysics , 7 , 158 metzger , b. d. , giannios , d. , & mimica , p. 2012
, , 420 , 3528 rees , m. j. 1988 , , 333 , 523 sambruna , r. m. , donato , d. , ajello , m. , maraschi , l. , & the gsfc bat team 2009 , arxiv:0912.2721 savolainen , t. , homan , d. c. , hovatta , t. , et al .
2010 , , 512 , a24 tang , q .- w . , peng , f .- k . ,
wang , x .- y . , & tam , p .- h .
t. 2015 , , 806 , 194 van velzen , s. , anderson , g. e. , stone , n. c. , et al .
2015 , arxiv:1511.08803 wang , f. y. , & cheng , k. s. 2012 , , 421 , 908 wang , j .- z . , lei , w .- h . , wang , d .- x . , et al .
2014 , , 788 , 32 wu , q. , zhang , b. , lei , w .- h . , et al .
2016 , , 455 , l1 zauderer , b. a. , berger , e. , soderberg , a. m. , et al .
2011 , , 476 , 425 -ray flare periods for relativistic sources : tdes , grbs and blazars .
sgrb and lgrb mean short grb and long grb respectively .
the three dished lines from left to right denote the 100% , 10% , and 1% fluence ratio between 0.1 - 10 gev and 10 - 1000 kev . ]
._fermi _ lat upper limit flux ( @xmath4 confidence level ) of four tidal disruption events at three time intervals after their trigger time .
flux is in unit of erg @xmath34 s@xmath3 ( 0.1 - 10 gev ) . [ cols="<,^,^,^,^,^,^,^,^ " , ] | massive black holes at galaxy center may tear apart a star when the star passes occasionally within the disruption radius , which is the so - called tidal disruption event(tde ) .
most tdes radiate with thermal emission resulted from the acceleration disk , but three tdes have been detected in bright non - thermal x - ray emission , which is interpreted as arising from the relativistic jets .
search for high - energy gamma - ray emission from one relativistic tde ( swift j164449.3 + 573451 ) with the _ fermi _ large area telescope ( lat ) has yielded non - detection . in this paper , we report the search for high energy emission from the other two relativistic tdes ( swift j2058.4 + 0516 swift j1112.2 - 8238 ) during the flare period .
no significant gev emission is found , with an upper limit fluence in lat energy range being less than @xmath0 of that in x - rays .
compared with gamma - ray bursts ( grbs ) and blazars , these tdes have the lowest flux ratio between gev emission and x - ray emission .
the non - detection of high - energy emission from relativistic tdes implies that the high - energy emission is heavily absorbed by soft photons in the source .
an upper limit on the bulk lorentz factor , @xmath1 , is then obtained for the jets in these tdes .
we also search for high - energy gamma - ray emission from the nearest tde discovered to date , asassn-14li .
no significant gev emission is found and an upper limit of @xmath2 erg s@xmath3 ( at @xmath4 confidence level ) is obtained for the first @xmath5 s after the disruption . |
optimization for multiple conflicting objectives results in more than one optimal solutions ( known as pareto - optimal solutions ) .
although one of these solutions is to be chosen at the end , the recent trend in evolutionary and classical multi - objective optimization studies have focused on approximating the set of pareto - optimal solutions . however , to assess the quality of pareto approximation set , special measures are needed @xcite .
hypervolume indicator is a commonly accepted quality measure for comparing approximation set generated by multi - objective optimizers .
the indicator measures the hypervolume of the dominated portion of the objective space by pareto approximation set and has received more and more attention in recent years @xcite .
there have been some studies that discuss the issue of fast hypervolume calculation @xcite .
these algorithms partition the covered space into many cuboid - shaped regions , within which the approach considering the dominated hypervolume as a special case of klee s measure problem is regarded as the current best one .
this approach @xcite adopts orthogonal partition tree which requires @xmath2 storage and streaming variant @xcite .
conceptual simplification of the implementation are concerned and thus the algorithm achieves an upper bound of @xmath6 for the hypervolume calculation . ignoring the running time of sorting the points according to the @xmath1-th dimension
, @xmath7 , the running time of this approach is exponential of the dimension of space @xmath1 .
this paper develops novel heuristics for the calculation of hypervolume indicator .
special technologies are applied and the novel approach yields upper bound of @xmath4 runtime and consumes @xmath5 storage .
the paper is organized as follows . in the next section
, the hypervolume indicator is defined , and some background on its calculation is provided . then , an algorithm is proposed which uses the so - called vertex - splitting technology to reduce the hypervolume .
the complexities of the proposed algorithm are analyzed in section [ sec : complexity ] .
the last section concludes this paper with an open problem .
without loss of generality , for multi - objective optimization problems , if the @xmath1 objective functions @xmath8 are considered with @xmath9 to be minimized , not one optimal solution but a set of good compromise solutions are obtained since that the objectives are commonly conflicting .
the compromise solutions are commonly called pareto approximation solutions and the set of them is called the pareto approximation set . for a pareto approximation set @xmath10 produced in a run of a multi - objective optimizer , where @xmath11 , all the solutions are non - comparable following the well - known concept of pareto dominance . specially , we say that @xmath12 dominates @xmath13 at the @xmath14-th dimension if @xmath15 .
the unary hypervolume indicator of a set @xmath16 consists of the measure of the region which is simultaneously dominated by @xmath16 and bounded above by a reference point @xmath17 such that @xmath18 . in the context of hypervolume indicator , we call the solutions in @xmath16 as the dominative points . as illustrated in fig .
[ fig:2d - set ] , the shading region consists of an orthogonal polytope , and may be seen as the union of three axis - aligned hyper - rectangles with one common vertex , i.e. , the reference point @xmath19 .
another example in three dimensional space is shown in fig .
[ fig:3d - set ] , where five dominative points , @xmath20 , and the reference point @xmath21 are considered .
the volume is the union of the volumes of all the cuboids each of which is bounded by a vertex , where the common regions are counted only once .
if a point @xmath13 is dominated by another point @xmath12 , the cuboid bounded by @xmath13 is completely covered by the cuboid bounded by @xmath12 . and thus only the non - dominated points contribute to the hypervolume .
in other works , e.g. the work of beume and rudolph @xcite , the hyper - cuboid in @xmath1-dimensional space are partitioned into child hyper - cuboids along the @xmath1-th dimension and then all these child hypervolumes are gathered together by the inclusion - exclusion principle @xcite . in this paper , we step in another way .
the hyper - cuboid is partitioned into child hyper - cuboids at some splitting reference points and then all the child hypervolumes are gathered directly .
more detailed , given a point @xmath22 , each of other points in @xmath16 must dominated @xmath12 at some dimensions for the non - comparable relation .
if the parts over @xmath12 are handled , the problem of calculating the hypervolume bounded by @xmath16 and the reference point is figured out .
the additional part partitioned out at the @xmath14-th dimension is also a @xmath1-dimensional hyper - cuboid whose vertices are ones beyond @xmath12 at such dimension .
their projections on the hyperplane orthogonal to dimension @xmath14 are all dominated by @xmath12 , and thus are free from consideration .
it should be noted that the reference point of child hyper - cuboid is altered to @xmath23 , namely the @xmath14-th coordinate is replaced by @xmath24 .
the other child hyper - cuboids are handled in the similar way . in these processes ,
the given point is called the splitting reference point . obviously
, the hyper - cuboids with more dominative points require more run time to calculate the hypervolumes . to reduce the whole run time for calculating all these child hyper - cuboids ,
the splitting reference point should be carefully selected .
the strategy adopted in this paper is described as follows .
\(1 ) let @xmath25 and choose a point with the least dimensions on which the point dominated by other @xmath26 points .
\(2 ) if some points tie , update @xmath26 as @xmath27 and then within these points , choose a point with the least dimensions on which the point dominated by other @xmath26 points .
\(3 ) repeat the similar process until only single point is left or @xmath28 . and
if @xmath28 and several points are left , the first found point is selected . by the above principle , as an example , not @xmath29 or other points but
@xmath30 is chosen as the first splitting reference point for the case shown in fig .
[ fig:3d - set ] .
two child cuboids each bounded by one points and another child cuboid bounded by two points are generated by splitting along @xmath30 .
this is the optimal strategy in such case .
the algorithm to calculate the hypervolume is shown in algorithm [ algo : calchypervolume ] .
some major parameters are as follows . *
* int[n][d ] order * the orders of all the dominative points at each dimension are represented by a two - dimensional array of integer .
* * int split * the index of the point at which the hyper - cuboid is cut to generate multiple child hyper - cuboids is called @xmath31 . *
* int[n ] splitcount * the numbers of @xmath26 present in the @xmath31-th row of the array @xmath32 are saved in @xmath33 , where @xmath34 .
* * int[n ] coveredcount * the numbers of @xmath26 present in the current checked row of the array @xmath32 are save in @xmath35 , where @xmath34 .
moreover , some conventions are explained as follows . *
the subscript of @xmath24 begins with @xmath36 while the index of array begins with @xmath37 .
thus @xmath24 is same as @xmath38[j-1]$ ] . *
assume @xmath39 and @xmath40 are two arrays and @xmath0 is an element .
@xmath41\leftarrow n$ ] means setting each element of @xmath39 as @xmath0 , while @xmath42 \leftarrow b [ ] $ ] means copying all the elements of @xmath40 to @xmath39 pairwise . *
assume @xmath43 is a set and @xmath44 is an element .
@xmath45 means appending a copy of @xmath44 to @xmath43 .
the inputs of the algorithm are a set of non - dominated ( dominative ) points and a reference point , thus the hyper - cuboids are represented implicitly .
the hyper - cuboid @xmath46 defined by the dominative points @xmath47 where @xmath48 , and the reference point @xmath49 , namely @xmath50 .
the initial number @xmath0 of dominative points can be obtained from the length of @xmath46 and the dimension @xmath1 is known too.the hypervolume of @xmath46 , @xmath51 .
@xmath52 ; @xmath53;@xmath54 \leftarrow n$ ] ; sort @xmath55;@xmath56[j-1 ] \leftarrow $ ] number of points dominating @xmath24 strictly ; @xmath57 \leftarrow 0 $ ] ; @xmath58$]++ ; @xmath59 ; @xmath54 \leftarrow coveredcount[]$ ] ; * break * ; @xmath60 ; @xmath61 ; @xmath62 ;
@xmath63 ; @xmath64 \leftarrow y_{split , j}$ ] ; @xmath65 ; @xmath66 ; @xmath67;@xmath51 ; in fact , when the hyper - cuboid is cut into two child hyper - cuboids , there may be some points dominated by the splitting reference point in the bigger cuboid , and thus such points could be removed from the points set @xmath46 . in the proposed algorithm , it does not matter whether those points are removed or not .
before discussing the time - space complexity of the proposed algorithm , some properties are presented firstly .
[ lemma : delta ] let @xmath68 be the number of points dominating @xmath12 at the @xmath14-th dimension .
then \(1 ) for @xmath69 and each @xmath70 , @xmath71 .
\(2 ) for @xmath69 and each @xmath72 , @xmath73 .
\(3 ) for @xmath74 and each @xmath70 , @xmath75 .
\(4 ) @xmath76 .
\(5 ) for @xmath69 and each @xmath70 , @xmath77 .
it is clear that ( 2 ) @xmath78 ( 4 ) @xmath78 ( 5 ) .
the follows show ( 1 ) , ( 2 ) and ( 3 ) .
\(1 ) ( by contradiction . )
assume to the contrary there is some @xmath79 , @xmath80 .
if this is the case , there are at least one @xmath13 where @xmath81 such that each @xmath24 dominates @xmath82 for all @xmath83 .
it follows that @xmath12 dominates @xmath13 , which contradicts our assumption that all the elements in @xmath84 are non - comparable .
\(2 ) given @xmath14 , sort all @xmath24 where @xmath85 and label each @xmath24 a sequence number @xmath86 which ranges from 0 to @xmath87 . thus @xmath88 .
there are two cases to consider .
firstly , if all @xmath24 are different each other , then @xmath89 .
it follows that @xmath90 . secondly , if there are same elements within @xmath91 , without loss of generality , suppose @xmath92 and @xmath93 .
then @xmath94 , it follows that @xmath95 .
this completes the proof .
\(3 ) ( by contradiction . ) for any @xmath12 , @xmath80 is excluded by ( 1 ) of this lemma .
thus @xmath96 for some @xmath12 is considered .
if this is the case , we obtain @xmath97 , contradicting ( 2 ) of this lemma , which implies @xmath98 , namely @xmath99 .
[ lemma : omega ] let @xmath100 be the amount of @xmath26 in all @xmath68 where @xmath101 , namely @xmath102 .
then \(1 ) @xmath103 for any @xmath104 and @xmath26 ; \(2 ) @xmath105 for any @xmath26 ; \(3 ) @xmath106 for any @xmath104 . by the definition of @xmath107 ,
it is clear that all statements follows lemma [ lemma : delta ] .
[ lemma : runtime ] let @xmath108 be the runtime of algorithm [ algo : calchypervolume ] to compute a hypervolume with @xmath0 dominative points in a @xmath1-dimensional space .
then \(1 ) @xmath109 where @xmath110 ; \(2 ) @xmath111 where @xmath112 ; \(3 ) @xmath108 is minimal when @xmath113 and @xmath114 for any @xmath14 and @xmath26 ; \(4 ) @xmath108 is maximal when @xmath115 for any @xmath104 and @xmath116 for any @xmath104 and each @xmath34 .
\(1 ) and ( 2 ) are clear .
\(3 ) by the process of algorithm [ algo : calchypervolume ] , given some @xmath104 , @xmath117 by ( 1 ) of lemma [ lemma : delta ] , @xmath71 .
it is clear that for a given @xmath104 , it is necessary that @xmath118 to minimize @xmath108 .
in addition , all the @xmath68 must share alike , i.e. @xmath114 for any @xmath14 and @xmath26 .
if this is not the truth , suppose @xmath119 .
thus by ( 1 ) of this lemma , @xmath120 let @xmath121 and @xmath122 .
@xmath123 and @xmath124 can be modified in the similar way until @xmath125 .
this completes the proof .
\(4 ) by ( 5 ) of lemma [ lemma : delta ] , @xmath126 .
it is clear that for a given @xmath104 , it is necessary that @xmath115 to maximize @xmath108 .
hence eqn .
( [ eqn : fnd1 ] ) is written as follows , @xmath127 suppose @xmath12 is the splitting reference point chosen by algorithm [ algo : calchypervolume ] , @xmath128 , or else contradicting @xmath129 . to maximize @xmath108 in eqn .
( [ eqn : fnd2 ] ) , let @xmath130 . similarly , we get @xmath131 , @xmath132 , @xmath133 , and so on .
it is exactly @xmath134 .
this completes the proof .
first of all , it is clear that @xmath135 . by ( 3 ) of lemma [ lemma : runtime ] , the algorithm performs best when each @xmath68 shares alike for the chosen @xmath104 . if @xmath136 , @xmath137 for any @xmath14 .
thus @xmath138 which implies @xmath139 . if @xmath140 , @xmath141 for any @xmath14 . in the rough ,
we get @xmath142 it can be obtained from eqn .
( [ eqn : lowerruntime ] ) that @xmath143 even when @xmath144 is relaxed to @xmath145 .
fredman and weide @xcite have shown that klee s measure problem has a lower bound of @xmath146 for arbitrary @xmath147 . just as beume and rudolph
@xcite have mentioned , although it is unknown what the lower bound for calculating the hypervolume is , it is definitely not harder than solving kmp because it is a special case of kmp . therefore , there is a gap between the lower bound of the proposed algorithm and the actual lower bound of calculating the hypervolume . in the average cases ,
suppose that for the given splitting reference point @xmath12 , @xmath148 .
meanwhile , each @xmath68 shares alike , i.e. @xmath149 .
thus , @xmath150 which implies the runtime of the proposed algorithm is @xmath151 at the given cases . by ( 2 ) of lemma [ lemma : runtime ] , @xmath152 for any @xmath153 . and by ( 4 ) of lemma [ lemma : runtime ] , at the worst cases , we have @xmath154 which implies that the proposed algorithm for computing the hypervolume bounded by @xmath0 points and a reference point in @xmath1-dimensional space has a runtime of @xmath4 at the worst cases .
let @xmath155 be the used storage by algorithm [ algo : calchypervolume ] . in the proposed algorithm ,
every child hypervolume is calculated one by one . since the storage can be reused after the former computation has been completed , @xmath155 is only related to the maximum usage of all the computations of child hypervolumes .
hence , @xmath156 thus the upper bound of space is as follows .
@xmath157 where @xmath158 .
it is easy to obtain an @xmath5 space upper bound for the proposed algorithm . combining the above analyses together
, we obtain the time - space complexity of the proposed algorithm .
the hypervolume of a hyper - cuboid bounded by @xmath0 non - comparable points and a reference point in @xmath1-dimensional space can be computed in time @xmath4 using @xmath5 storage .
a fast algorithm to calculate the hypervolume indicator of pareto approximation set is proposed . in the novel algorithm ,
the hyper - cuboid bounded by non - comparable points and the reference point is partitioned into many child hyper - cuboids along the carefully chosen splitting reference point at each dimension .
the proposed approach is very different to the technique used in other works where the whole @xmath1-dimensional volume is calculated by computing the @xmath159-dimensional volume along the dimension @xmath1 .
such difference results in very different time bounds , namely @xmath4 for our work and @xmath160 for the best previous result .
neither kind of technique can exceed the other completely and each has his strong point .
additionally , the amount of storage used by our algorithm is only @xmath5 even no special technique is developed to reduce the space complexity .
as the context has mentioned , it is very important to choose appropriate splitting reference point for our algorithm .
well selected point can reduce number of points in separated parts and thus cut down the whole runtime .
we do not know whether the strategy adopted in this paper is optimal or near optimal .
further investigations should be worked on .
zitzler , e. , thiele , l. , laumanns , m. , fonseca , c.m . ,
da fonseca , v.g .
: performance assessment of multiobjective optimizers : an analysis and review .
ieee transactions on evolutionary computation * 7*(2 ) ( 2003 ) 117132 zitzler , e. , thiele , l. : multiobjective optimization using evolutionary algorithms a comparative study . in eiben , a.e . , ed . :
parallel problem solving from nature v , amsterdam , springer - verlag ( 1998 ) 292301 zitzler , e. , brockhoff , d. , thiele , l. : the hypervolume indicator revisited : on the design of pareto - compliant indicators via weighted integration . in : proceedings of the 4th international conference on evolutionary multi - criterion optimization ( emo 2007 ) .
volume 4403 .
, springer - verlag ( 2007 ) 862876 while , l. , bradstreet , l. , barone , l. , hingston , p. : heuristics for optimising the calculation of hypervolume for multi - objective optimisation problems . in : the 2005 ieee congress on evolutionary computation
. volume 3 .
( 2005 ) 22252232 beume , n. , rudolph , g. : faster s - metric calculation by considering dominated hypervolume as klee s measure problem . in kovalerchuk ,
b. , ed . : proceedings of the second iasted conference on computational intelligence , anaheim , acta press ( 2006 ) 231236 | hypervolume indicator is a commonly accepted quality measure for comparing pareto approximation set generated by multi - objective optimizers .
the best known algorithm to calculate it for @xmath0 points in @xmath1-dimensional space has a run time of @xmath2 with special data structures .
this paper presents a recursive , vertex - splitting algorithm for calculating the hypervolume indicator of a set of @xmath0 non - comparable points in @xmath3 dimensions .
it splits out multiple child hyper - cuboids which can not be dominated by a splitting reference point . in special , the splitting reference point
is carefully chosen to minimize the number of points in the child hyper - cuboids .
the complexity analysis shows that the proposed algorithm achieves @xmath4 time and @xmath5 space complexity in the worst case . |
ngc 5286 ( c1343 - 511 ) is a fairly bright ( @xmath6 ) and dense globular cluster ( gc ) , with a central luminosity density @xmath7 which is more than a factor of six higher than in the case of @xmath8 centauri ( ngc 5139 ) , according to the entries in the @xcite catalog . in ( * ? ? ?
* hereafter paper i ) we presented a color - magnitude diagram ( cmd ) study of the cluster that reveals an unusual horizontal branch ( hb ) morphology in that it does not contain a prominent red hb component , contrary to what is normally found in gcs with comparable metallicity ( @xmath2 } = -1.67 $ ] ; @xcite @xcite ) , such as m3 ( ngc 5272 ) or m5 ( ngc 5904 ) . as a matter of fact , ngc 5286 contains blue hb stars reaching down all the way to at least the main sequence turnoff level in @xmath9 . yet , unlike most blue hb gcs , ngc 5286 is known to contain a sizeable population of rr lyrae variable stars , with at least 15 such variables being known in the field of the cluster @xcite . in this sense ,
ngc 5286 resembles the case of m62 ( ngc 6266 ; * ? ? ?
* ) , thus possibly being yet another member of a new group of gcs with hb types intermediate between m13 ( ngc 6205)-like ( a very blue hb with relatively few rr lyrae variables ) and that of the oosterhoff i ( oo i ) cluster m3 ( a redder hb , with a well - populated instability strip ) . ngc 5286 thus constitutes an example of the `` missing link '' between m3- and m13-like gcs @xcite .
previous surveys for variable stars in ngc 5286 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) have turned up relatively large numbers of rr lyrae stars .
however , such studies were carried out either by photographic methods , used comparatively few observations , or utilized reduction methods that have subsequently been superseded by improved techniques , including robust multiple - frame photometry ( e.g. , allframe ; * ? ? ?
* ) and image subtraction ( e.g. , isis ; * ? ? ?
this , together with the large central surface brightness of the cluster , strongly suggests that a large population of variable stars remains unknown in ngc 5286 , especially towards its crowded inner regions .
in addition , for the known or suspected variables , it should be possible to obtain light curves of much superior quality to those available , thus leading to better defined periods , amplitudes , and fourier decomposition parameters .
indeed , to our knowledge , no modern variability study has ever been carried out for this cluster .
a study of its variable star population appears especially interesting in view of its suggested association with the canis major dwarf spheroidal galaxy @xcite , and the constraints that the ancient rr lyrae variable stars are able to pose on the early formation history of galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , the time seems ripe for a reassessment of the variable star content of ngc 5286 and this is precisely the main subject of the present paper .
in [ sec : obs ] , we describe the variable stars search techniques and the conversion from isis relative fluxes to standard magnitudes . in [ sec : var ] , we show the results of our variability search , giving the positions , periods , amplitudes , magnitudes , and colors for the detected variables . we show the positions of the variables in the cluster cmd in [ sec : cmd ] . in [ sec : four ] , we provide the results of a fourier decomposition of the rr lyrae light curves , obtaining several useful physical parameters .
we analyze the cluster s oosterhoff type in [ sec : oos ] , whereas [ sec : cep ] is dedicated to the type ii cepheid that we found in ngc 5286 .
[ sec : summ ] summarizes the main results of our investigation .
all of the derived light curves are provided in an appendix .
lccccccccccccl v01 ........... & @xmath10 & @xmath11 & @xmath12 & @xmath13 & @xmath14 & @xmath15 & @xmath16 & @xmath17 & @xmath18 & @xmath19 & @xmath20 & @xmath21 & rr_ab _
+ v02 ........... & @xmath22 & @xmath23 & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 & @xmath32 & @xmath33 & rr_ab _
+ v03 ........... & @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & rr_ab _
+ v04 ........... & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & @xmath56 & @xmath57 & rr_c _
+ v05 ........... & @xmath58 & @xmath59 & @xmath60 & @xmath61 & @xmath62 & @xmath61 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & @xmath66 & @xmath66 & rr_ab _
+ v06 ........... & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & rr_ab _
+ v07 ........... & @xmath79 & @xmath80 & @xmath81 & @xmath61 & @xmath61 & @xmath82 & @xmath83 & @xmath84 & @xmath84 & @xmath85 & @xmath86 & @xmath86 & rr_ab _
+ v08 ...... & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath96 & @xmath66 & bl her + v09 ........... & @xmath97 & @xmath98 & @xmath99 & @xmath100 & @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & @xmath108 & rr_c _
+ v10 ......... & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & rr_ab _
+ v11 ......... & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & @xmath128 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & rr_ab _
+ v12 ......... & @xmath133 & @xmath134 & @xmath135 & @xmath136 & @xmath101 & @xmath137 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 & @xmath143 & rr_c _
+ v13 ......... & @xmath58 & @xmath144 & @xmath145 & @xmath136 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 & rr_c _
+ v14 ......... & @xmath154 & @xmath155 & @xmath156 & @xmath49 & @xmath157 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & @xmath164 & rr_c _
+ v15 ......... & @xmath165 & @xmath166 & @xmath167 & @xmath168 & @xmath169 & @xmath17 & @xmath170 & @xmath171 & @xmath149 & @xmath172 & @xmath173 & @xmath174 & rr_ab _ + v17 ......... & @xmath175 & @xmath176 & @xmath177 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & @xmath184 & @xmath185 & @xmath186 & rr_ab _
+ v18 ......... & @xmath187 & @xmath188 & @xmath189 & @xmath190 & @xmath61 & @xmath191 & @xmath61 & @xmath192 & @xmath61 & @xmath61 & @xmath61 & @xmath61 & rr_ab _
+ v20 ......... & @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 & @xmath199 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & @xmath203 & rr_c _
+ v21 ......... & @xmath204 & @xmath205 & @xmath69 & @xmath206 & @xmath38 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & @xmath143 & @xmath212 & rr_ab _ + nv1 ........ & @xmath213 & @xmath214 & @xmath215 & @xmath216 & @xmath217 & @xmath218 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & @xmath78 & @xmath223 & rr_c _
+ nv2 ........ & @xmath224 & @xmath225 & @xmath226 & @xmath49 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 & @xmath234 & rr_c _
+ nv3 ........ & @xmath235 & @xmath236 & @xmath237 & @xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & @xmath245 & @xmath246 & rr_ab _ + nv4 ........ & @xmath247 & @xmath248 & @xmath249 & @xmath250 & @xmath251 & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath258 & rr_ab _
+ nv5 ........ & @xmath259 & @xmath260 & @xmath261 & @xmath49 & @xmath262 & @xmath263 & @xmath264 & @xmath265 & @xmath266 & @xmath33 & @xmath267 & @xmath268 & rr_c _ + nv6 ........ & @xmath269 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath215 & rr_ab _ + nv7 ........ & @xmath204 & @xmath280 & @xmath281 & @xmath101 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 & @xmath288 & @xmath289 & rr_c _ + nv8 ........ & @xmath121 & @xmath290 & @xmath291 & @xmath239 & @xmath239 & @xmath292 & @xmath293 & @xmath294 & @xmath295 & @xmath296 & @xmath296 & @xmath256 & rr_ab _ + nv9 ........ & @xmath297 & @xmath298 & @xmath299 & @xmath300 & @xmath300 & @xmath301 & @xmath83 & @xmath64 & @xmath300 & @xmath302 & @xmath300 & @xmath61 & rr_ab _
+ nv10 ...... & @xmath303 & @xmath304 & @xmath281 & @xmath305 & @xmath216 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 & @xmath311 & @xmath312 & rr_c _
+ nv11 ...... & @xmath313 & @xmath314 & @xmath315 & @xmath316 & @xmath26 & @xmath317 & @xmath318 & @xmath319 & @xmath320 & @xmath321 & @xmath76 & @xmath322 & rr_ab _
+ nv12 ...... & @xmath323 & @xmath324 & @xmath325 & @xmath326 & @xmath282 & @xmath327 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & @xmath332 & @xmath333 & rr_ab _ + nv13 ...... & @xmath334 & @xmath335 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & @xmath340 & @xmath286 & @xmath341 & @xmath288 & @xmath342 & @xmath343 & rr_ab _ + nv14 ...... & @xmath344 & @xmath345 & @xmath346 & @xmath347 & @xmath348 & @xmath349 & @xmath140 & @xmath350 & @xmath351 & @xmath352 & @xmath353 & @xmath354 & rr_c _ + nv15 ...... & @xmath355 & @xmath356 & @xmath357 & @xmath358 & @xmath326 & @xmath359 & @xmath360 & @xmath361 & @xmath362 & @xmath363 & @xmath364 & @xmath365 & rr_ab _ + nv16 ...... & @xmath366 & @xmath367 & @xmath215 & @xmath216 & @xmath368 & @xmath369 & @xmath370 & @xmath371 & @xmath372 & @xmath33 & @xmath311 & @xmath373 & rr_c _
+ nv17 ...... & @xmath374 & @xmath375 & @xmath376 & @xmath146 & @xmath227 & @xmath377 & @xmath378 & @xmath379 & @xmath380 & @xmath381 & @xmath156 & @xmath381 & rr_c _
+ nv18 ...... & @xmath382 & @xmath383 & @xmath384 & @xmath385 & @xmath348 & @xmath386 & @xmath387 & @xmath388 & @xmath380 & @xmath389 & @xmath390 & @xmath391 & rr_c _ + nv19 ...... & @xmath392 & @xmath393 & @xmath394 & @xmath124 & @xmath395 & @xmath396 & @xmath397 & @xmath398 & @xmath399 & @xmath400 & @xmath401 & @xmath402 & rr_ab _ + nv20 ...... & @xmath403 & @xmath404 & @xmath405 & @xmath406 & @xmath136 & @xmath407 & @xmath408 & @xmath409 & @xmath410 & @xmath411 & @xmath412 & @xmath413 & rr_c _
+ nv21 ...... & @xmath79 & @xmath414 & @xmath415 & @xmath337 & @xmath416 & @xmath417 & @xmath418 & @xmath396 & @xmath419 & @xmath420 & @xmath421 & @xmath422 & rr_ab _ + nv22 ...... & @xmath423 & @xmath424 & @xmath305 & @xmath91 & @xmath425 & @xmath426 & @xmath427 & @xmath428 & @xmath429 & @xmath430 & @xmath431 & @xmath432 & rr_ab _
+ nv23 ...... & @xmath323 & @xmath433 & @xmath185 & @xmath434 & @xmath435 & @xmath436 & @xmath437 & @xmath182 & @xmath438 & @xmath439 & @xmath440 & @xmath441 & rr_ab _
+ nv24 ...... & @xmath382 & @xmath442 & @xmath49 & @xmath443 & @xmath444 & @xmath445 & @xmath446 & @xmath447 & @xmath448 & @xmath449 & @xmath336 & @xmath450 & rr_ab _ + nv25 ...... & @xmath451 & @xmath452 & @xmath142 & @xmath453 & @xmath425 & @xmath454 & @xmath455 & @xmath456 & @xmath457 & @xmath458 & @xmath459 & @xmath460 & rr_ab _ + nv26 ...... & @xmath313 & @xmath461 & @xmath462 & @xmath50 & @xmath463 & @xmath464 & @xmath465 & @xmath466 & @xmath467 & @xmath468 & @xmath373 & @xmath469 & rr_c _ + nv27 ...... & @xmath470 & @xmath471 & @xmath472 & @xmath473 & @xmath291 & @xmath474 & @xmath475 & @xmath476 & @xmath477 & @xmath469 & @xmath32 & @xmath478 & rr_ab _
+ nv28 ...... & @xmath479 & @xmath480 & @xmath353 & @xmath481 & @xmath444 & @xmath482 & @xmath483 & @xmath484 & @xmath485 & @xmath469 & @xmath32 & @xmath486 & rr_ab _ + nv29 ...... & @xmath487 & @xmath488 & @xmath489 & @xmath481 & @xmath100 & @xmath490 & @xmath491 & @xmath492 & @xmath493 & @xmath494 & @xmath44 & @xmath495 & rr_c _ + nv30 ...... & @xmath382 & @xmath496 & @xmath497 & @xmath136 & @xmath238 & @xmath498 & @xmath499 & @xmath240 & @xmath500 & @xmath501 & @xmath502 & @xmath458 & rr_ab _ + nv31 ...... & @xmath355 & @xmath503 & @xmath504 & @xmath282 & @xmath505 & @xmath506 & @xmath129 & @xmath507 & @xmath159 & @xmath508 & @xmath509 & @xmath431 & rr_c _ + nv32 ...... & @xmath193 & @xmath510 & @xmath511 & @xmath358 & @xmath512 & @xmath513 & @xmath514 & @xmath515 & @xmath516 & @xmath517 & @xmath495 & @xmath96 & rr_c _ + nv33 ...... & @xmath518 & @xmath519 & @xmath145 & @xmath358 & @xmath227 & @xmath520 & @xmath521 & @xmath522 & @xmath523 & @xmath524 & @xmath525 & @xmath526 & rr_c _ + nv34 ...... & @xmath323 & @xmath527 & @xmath528 & @xmath529 & @xmath101 & @xmath530 & @xmath531 & @xmath532 & @xmath533 & @xmath534 & @xmath494 & @xmath535 & rr_c _ + nv35 ...... & @xmath536 & @xmath537 & @xmath61 & @xmath61 & @xmath61 & @xmath538 & @xmath539 & @xmath61 & @xmath61 & @xmath540 & @xmath61 & @xmath61 & lpv + nv36 ...... & @xmath382 & @xmath541 & @xmath61 & @xmath61 & @xmath61 & @xmath542 & @xmath543 & @xmath61 & @xmath61 & @xmath544 & @xmath61 & @xmath61 & lpv + nv37 ...... & @xmath487 & @xmath545 & @xmath61 & @xmath61 & @xmath61 & @xmath546 & @xmath547 & @xmath61 & @xmath61 & @xmath548 & @xmath61 & @xmath61 & lpv + nv38 ...... & @xmath487 & @xmath549 & @xmath61 & @xmath61 & @xmath61 & @xmath542 & @xmath550 & @xmath61 & @xmath61 & @xmath551 & @xmath61 & @xmath61 & lpv [ tab1 ]
the images used in this paper are the same as described in paper i , constituting a set of 128 frames in @xmath9 and 133 in @xmath552 , acquired with the 1.3 m warsaw university telescope at las campanas observatory , chile , in the course of a one - week run in april 2003 .
further details can be found in paper i. in addition , a few images were taken in feb .
2008 using the 4.1 m southern astrophysical research ( soar ) telescope , located in cerro pachn , chile , to further check the positions of the variables in the crowded regions around the cluster center .
the variable stars search was made using the image subtraction package isis v2.2 @xcite . in order to convert the isis differential fluxes to standard magnitudes , we used daophot ii / allframe @xcite to obtain instrumental magnitudes for each of the variables in the @xmath552 and @xmath9 reference images of the isis reductions .
first we obtained the flux of the variable star in the reference image , given by @xmath553 where @xmath554 is the instrumental magnitude of the star in the reference image and @xmath555 is a constant which depends on the photometric reduction package ( for daophot ii / allframe it is @xmath555 = 25 ) .
then we derived instrumental magnitudes for each epoch from the differential fluxes @xmath556 given by isis using the equation @xmath557 finally , the equations to obtain the calibrated magnitudes ( @xmath558 ) from the instrumental magnitudes are of the following form : @xmath559 where @xmath560 is the calibrated magnitude of the star in the reference image ( we used the standard magnitude data from paper i ) . lccccc v01 & @xmath9 & @xmath561 & @xmath562 & @xmath563 & @xmath564 + v01 & @xmath9 & @xmath565 & @xmath566 & @xmath567 & @xmath568 + v01 & @xmath9 & @xmath569 & @xmath570 & @xmath571 & @xmath572 + v01 & @xmath9 & @xmath573 & @xmath574 & @xmath575 & @xmath576 + v01 & @xmath9 & @xmath577 & @xmath578 & @xmath579 & @xmath580 + v01 & @xmath9 & @xmath581 & @xmath582 & @xmath583 & @xmath580 + [ longtab ]
in our variability search , we found 57 variable stars : 52 rr lyrae ( 22 rr_c _ , 30 rr_ab _ ) , 4 lpv s , and 1 type ii cepheid ( more specifically , a bl herculis star ) .
we identified 19 of the 24 previously catalogued variables , and discovered 38 new variables .
a finding chart is provided in figure [ ubica ] .
of the 16 previously catalogued variables with known periods ( v1-v16 ) we were able to find 15 .
v16 is the only one not present in our data , because it is not in the chip of the ccd that we have analyzed for variability .
the other 8 previously catalogued variables ( v17-v24 ) were suggested by @xcite based just on their position on the cmd .
we found that only 4 of these stars ( v17 , v18 , v20 , and v21 ) are real variables in our survey .
we do not detect any variable sources at the coordinates that they provide for the remaining 4 candidate variables in their study ( v19 , v22 , v23 and v24 ) .
the recent images taken with a better spatial resolution at the 4.1 m soar telescope reveal that v19 is very close to two other stars , and probably is not resolved in the images used by @xcite .
v22 and v23 are close to the instability strip but they still fall in the blue part of the hb , so they are not variable stars .
v24 is in the instability strip but very close to the blue part of the hb .
it is possible that this star belongs to the blue hb and is contaminated by a redder star .
periods were determined using the phase dispersion minimization ( pdm ; @xcite ) program in iraf .
periods , along with the coordinates and several important photometric parameters , are provided in table [ tab1 ] . in this table ,
column 1 indicates the star s name .
columns 2 and 3 provide the right ascension and declination ( j2000 epoch ) , respectively , whereas column 4 shows our derived period .
columns 5 and 6 list the derived amplitudes in the @xmath552 and @xmath9 bands , respectively , whereas columns 7 and 8 show the magnitude - weighted mean @xmath552 and @xmath9 magnitudes , corrected for differential reddening ( see paper i for details ) .
the corresponding intensity - mean averages are provided in columns 9 and 10 ( also corrected for differential reddening ) .
the average @xmath584 color in magnitude units and the intensity - mean color @xmath585 are given in columns 11 and 12 , respectively , whereas column 13 lists the @xmath584 color corresponding to the equivalent static star .
finally , the last column indicates the star s variability type . to derive the color of the equivalent static star ( i.e. , the color the star would have if it were not pulsating ) , we first derived the magnitude - weighted mean color , and then applied an amplitude - dependent correction by interpolating on table 4 from @xcite .
we calculate the hb level , @xmath586 , as the average @xmath587 magnitude of all the rr lyrae detected .
magnitude data as a function of julian date and phase for the variable stars detected in our study are given in table [ longtab ] . in this table ,
column 1 indicates the star s name , following the @xcite designation ( when available ) .
column 2 indicates the filter used .
column 3 provides the julian date of the observation , whereas column 4 shows the phase according to our derived period ( from table [ tab1 ] ) .
columns 5 and 6 list the observed magnitude in the corresponding filter and the associated error , respectively .
light curves based on our derived periods ( when available ) are shown in the appendix .
figure [ cmdcv ] shows the variable stars in the cmd , decontaminated from field stars as described in paper i. we can see that all rr lyrae stars fall around the hb region , whereas the type ii cepheid is brighter than the hb .
the four detected lpvs all fall close to the top of the rgb .
these trends are precisely as expected if all the detected variables are cluster members .
note , however , that the cmd positions of the 4 lpvs are not precisely defined , since we lack adequate phase coverage for these stars .
figure [ hb ] is a magnified cmd showing only the hb region . to assess the effects of crowding , we use different symbol sizes for the variables in different radial annuli from the cluster center : small sizes for stars in the innermost cluster regions ( @xmath588 ) which are badly affected by crowding , medium sizes for stars with @xmath589 , and large sizes for stars in the outermost cluster regions ( @xmath590 ) .
as expected , the variables in the innermost cluster region present more scatter . apart from this
, in general the detected rr lyrae fall inside a reasonably well - defined instability strip ( is ) .
however , we can see that there is not a clear - cut separation in color between rr_c _ s and rr_ab _ s .
although the rr_c _ s tend to be found preferentially towards the blue side of the is , as expected , the rr_ab _ s are more homogeneously distributed .
this can be again a scatter effect , because the rrab stars that are in the less crowded regions of the cluster ( large and medium size circles ) are more concentrated at the red part . * v5 * : this star was only detected in our @xmath9 images , so that we were unable to determine its color .
it remains unclear to us why isis was unable to detect this variable in the @xmath552 data note , from figure [ ubica ] , that it is not located in an especially crowded region of the cluster .
* v7 * : our derived period , 0.512 d , is slightly longer than the previously reported period of 0.50667 d in @xcite . as a matter of fact
, v7 remains one of the shortest - period rrab stars among all known ooii gcs . because the period is so close to half a day
, there is a considerable gap in the phased light curve for v7 ( as was also the case , though to a somewhat lesser degree , for the light curve of @xcite ) .
we carefully checked the pdm periodogram of the star in search of acceptable longer periods , but could find none that fit our data , nor could we find a period that reduced significantly the spread seen in the phased light curve close to minimum light ( which is particularly obvious in the @xmath9-band light curve ) .
the @xcite light curve also shows substantial scatter close to minimum light , although in their case much of the scatter is clearly due to photometric error . *
v8 * : @xcite found a period of 0.7 d for this star and suggested that it is an rr lyrae - type variable . in our analysis we found an alias at 0.7 d for the period , but the best fit is obtained with a period of 2.33 d , which corresponds to a type ii cepheid of the bl her type . * v18 * : as for v5 , we only detected this star in one of the filters . in this case
isis only detected it in our @xmath552 images , so that we were unable to determine its color and also to perform fourier decomposition .
* v7 * and * nv9 * : we do not see the minimum and maximum , respectively , for these two variables .
for that reason , we could not perform a fourier analysis , and the mean @xmath9 and @xmath552 values provided in table [ tab1 ] are just approximate .
light curves for rr lyrae variables were analyzed by fourier decomposition using the same equations as in @xcite , namely @xmath591 ( for rr_ab _ variables ) and @xmath592 ( for rr_c _ variables ) , where again @xmath593 was usually adopted .
amplitude ratios @xmath594 and phase differences @xmath595 for the lower - order terms are provided in table [ tab2 ] . in this table , a colon symbol ( `` : '' ) indicates an uncertain value , whereas a double colon ( `` : : '' ) indicates a very uncertain value .
lllllll v4 & @xmath596 & @xmath597 & @xmath598 & @xmath599 & @xmath600 & @xmath601 + v9 & @xmath602 & @xmath603 & @xmath604 & @xmath605 & @xmath606 & @xmath607 + v12 & @xmath608 & @xmath609 & @xmath610 & @xmath611 & @xmath612 & @xmath613 + v13 & @xmath614 & @xmath615 & @xmath616 & @xmath617 & @xmath618 & @xmath619 + v14 & @xmath620 & @xmath621 & @xmath622 & @xmath623 & @xmath624 & @xmath625 + v20 & @xmath626 & @xmath627 & @xmath628 & @xmath629 & @xmath630 & @xmath631 + nv1 & @xmath632 & @xmath633 & @xmath604 & @xmath634 & @xmath635 & @xmath636 + nv2 & @xmath637 & @xmath638 & @xmath639 & @xmath640 & @xmath641 & @xmath642 + nv5 & @xmath643 & @xmath597 & @xmath644 & @xmath645 & @xmath646 & @xmath647 + nv7 & @xmath648 & @xmath649 & @xmath650 & @xmath651 & @xmath652 & @xmath653 + nv10 & @xmath654 & @xmath603 & @xmath655 & @xmath656 & @xmath657 & @xmath658 + nv14 & @xmath659 & @xmath604 & @xmath660 & @xmath661 & @xmath662 & @xmath663 + nv16 & @xmath664 & @xmath665 & @xmath666 & @xmath667 & @xmath668 & @xmath669 + nv17 & @xmath670 & @xmath671 & @xmath672 & @xmath673 & @xmath674 & @xmath675 + nv18 & @xmath676 & @xmath677 & @xmath678 & @xmath679 & @xmath680 & @xmath681 + nv20 & @xmath682 & @xmath683 & @xmath684 & @xmath685 & @xmath686 & @xmath687 + nv26 & @xmath688 & @xmath689 & @xmath690 & @xmath691 & @xmath692 & @xmath693 + nv29 & @xmath694 & @xmath695 & @xmath696 & @xmath697 & @xmath698 & @xmath699 + nv31 & @xmath700 & @xmath701 & @xmath702 & @xmath703 & @xmath704 & @xmath705 + nv32 & @xmath706 & @xmath707 & @xmath708 & @xmath709 & @xmath710 & @xmath711 + nv33 & @xmath712 & @xmath713 & @xmath714 & @xmath715 & @xmath716 & @xmath717 + nv34 & @xmath718 & @xmath719 & @xmath707 & @xmath720 & @xmath721 & @xmath722 [ tab2 ] lcccccc v4 & @xmath723 & @xmath724 & @xmath725 & @xmath726 & @xmath727 & @xmath728 + v9 & @xmath729 & @xmath730 & @xmath731 & @xmath732 & @xmath733 & @xmath734 + v12 & @xmath185 & @xmath735 & @xmath736 & @xmath737 & @xmath738 & @xmath739 + v13 & @xmath331 & @xmath740 & @xmath741 & @xmath742 & @xmath743 & @xmath744 + v14 & @xmath494 & @xmath745 & @xmath746 & @xmath402 & @xmath747 & @xmath748 + v20 & @xmath749 & @xmath750 & @xmath751 & @xmath504 & @xmath752 & @xmath753 + nv1 & @xmath754 & @xmath755 & @xmath756 & @xmath757 & @xmath758 & @xmath299 + nv2 & @xmath759 & @xmath735 & @xmath760 & @xmath737 & @xmath761 & @xmath357 + nv5 & @xmath77 & @xmath762 & @xmath763 & @xmath764 & @xmath765 & @xmath766 + nv10 & @xmath767 & @xmath768 & @xmath769 & @xmath770 & @xmath771 & @xmath772 + nv14 & @xmath773 & @xmath774 & @xmath775 & @xmath776 & @xmath777 & @xmath778 + nv20 & @xmath779 & @xmath780 & @xmath781 & @xmath732 & @xmath782 & @xmath783 + mean & @xmath784 & @xmath785 & @xmath786 & @xmath787 & @xmath788 & @xmath789 [ tab3 ] @xcite used light curves of rr_c _
variables , as provided by their hydrodynamical pulsation models , to derive equations to calculate mass , luminosity , effective temperature , and a `` helium parameter '' for rr_c _ variables all as a function of the fourier phase difference @xmath790 and the period .
however , ( * ? ? ?
* see his 4 ) pointed out that the @xcite set of equations can not , in their current form , provide physically correct values for luminosities and masses , since they violate the constraints imposed by the ritter ( period - mean density ) relation .
we still provide the derived values for the ngc 5286 rr_c _ stars in this paper though , chiefly for comparison with similar work for other gcs .
accordingly , we use @xcite s ( @xcite ) equations ( 2 ) , ( 3 ) , ( 6 ) , and ( 7 ) to compute @xmath791 , @xmath792 , @xmath793 , and _
y _ , respectively , for 12 of our rr_c _ variables ( i.e. , those with the best defined fourier coefficients )
. we also use equation ( 3 ) in @xcite to compute @xmath2}$ ] , and equation ( 10 ) in @xcite to compute @xmath794 .
the results are given in table [ tab3 ] .
the unweighted mean values and corresponding standard errors of the derived mean mass , @xmath795 luminosity ( in solar units ) , effective temperature , `` helium parameter '' , metallicity ( in the @xcite @xcite scale ) , and mean absolute magnitude in @xmath9 are @xmath796 , @xmath797 , @xmath798 k , @xmath799 , @xmath800 , and @xmath801 mag , respectively .
amplitude ratios @xmath802 and phase differences @xmath803 for the lower - order terms are provided , in the case of the rr_ab _
s , in table [ tab4 ] .
we also give the jurcsik - kovcs @xmath804 value ( @xcite , computed on the basis of their eq .
[ 6 ] and table 6 ) , which is intended to differentiate rr_ab _ stars with `` regular '' light curves from those with `` anomalous '' light curves ( e.g. , presenting the blazhko effect but see also @xcite @xcite for a critical discussion of @xmath804 as an indicator of the occurrence of the blazhko phenomenon ) . as before , a colon symbol ( `` : '' ) indicates an uncertain value , whereas a double colon ( `` : : '' ) indicates a very uncertain value .
lllllllr v1 & @xmath805 & @xmath806 & @xmath807 & @xmath808 & @xmath809 & @xmath810 & @xmath811 + v2 & @xmath812 & @xmath813 & @xmath814 & @xmath815 & @xmath816 & @xmath817 & @xmath818 + v3 & @xmath819 & @xmath820 & @xmath821 & @xmath822 & @xmath823 & @xmath824 & @xmath825 + v5 & @xmath826 & @xmath827 & @xmath828 & @xmath829 & @xmath830 & @xmath831 & @xmath832 + v6 & @xmath833 & @xmath834 & @xmath835 & @xmath836 & @xmath837 & @xmath838 & @xmath839 + v7 & @xmath840 & @xmath841 & @xmath841 & @xmath841 & @xmath841 & @xmath841 & @xmath841 + v10 & @xmath842 & @xmath843 & @xmath844 & @xmath845 & @xmath846 & @xmath847 & @xmath848 + v11 & @xmath849 & @xmath850 & @xmath851 & @xmath852 & @xmath853 & @xmath854 & @xmath855 + v15 & @xmath856 & @xmath857 & @xmath858 & @xmath859 & @xmath860 & @xmath861 & @xmath862 + v17 & @xmath863 & @xmath864 & @xmath865 & @xmath866 & @xmath867 & @xmath868 & @xmath869 + v21 & @xmath870 & @xmath871 & @xmath872 & @xmath873 & @xmath874 & @xmath875 & @xmath876 + nv3 & @xmath877 & @xmath878 & @xmath879 & @xmath880 & @xmath881 & @xmath882 & @xmath883 + nv4 & @xmath884 & @xmath885 & @xmath886 & @xmath887 & @xmath888 & @xmath889 & @xmath890 + nv6 & @xmath891 & @xmath892 & @xmath893 & @xmath894 & @xmath895 & @xmath896 & @xmath848 + nv8 & @xmath897 & @xmath898 & @xmath899 & @xmath900 & @xmath901 & @xmath902 & @xmath903 + nv9 & @xmath840 & @xmath841 & @xmath841 & @xmath841 & @xmath841 & @xmath841 & @xmath841 + nv11 & @xmath904 & @xmath905 & @xmath906 & @xmath907 & @xmath908 & @xmath909 & @xmath910 + nv12 & @xmath911 & @xmath912 & @xmath913 & @xmath914 & @xmath915 & @xmath916 & @xmath917 + nv13 & @xmath918 & @xmath919 & @xmath920 & @xmath921 & @xmath922 & @xmath923 & @xmath924 + nv15 & @xmath925 & @xmath926 & @xmath864 & @xmath927 & @xmath928 & @xmath929 & @xmath930 + nv19 & @xmath931 & @xmath828 & @xmath932 & @xmath933 & @xmath934 & @xmath935 & @xmath936 + nv21 & @xmath937 & @xmath938 & @xmath939 & @xmath940 & @xmath941 & @xmath942 & @xmath943 + nv22 & @xmath944 & @xmath945 & @xmath920 & @xmath946 & @xmath947 & @xmath948 & @xmath949 + nv23 & @xmath950 & @xmath951 & @xmath952 & @xmath953 & @xmath954 & @xmath955 & @xmath956 + nv24 & @xmath957 & @xmath958 & @xmath959 & @xmath960 & @xmath961 & @xmath962 & @xmath963 + nv25 & @xmath856 & @xmath964 & @xmath965 & @xmath966 & @xmath967 & @xmath968 & @xmath969 + nv27 & @xmath970 & @xmath971 & @xmath972 & @xmath973 & @xmath974 & @xmath975 & @xmath976 + nv28 & @xmath977 & @xmath978 & @xmath979 & @xmath980 & @xmath981 & @xmath982 & @xmath983 + nv30 & @xmath984 & @xmath985 & @xmath986 & @xmath987 & @xmath988 & @xmath989 & @xmath990 [ tab4 ] lcccccccc v1 & @xmath991 & @xmath992 & @xmath993 & @xmath994 & @xmath995 & @xmath996 & @xmath997 & @xmath998 + v2 & @xmath999 & @xmath1000 & @xmath1001 & @xmath1002 & @xmath995 & @xmath1003 & @xmath1004 & @xmath1005 + v3 & @xmath1006 & @xmath19 & @xmath1007 & @xmath1008 & @xmath215 & @xmath994 & @xmath1009 & @xmath1010 + v10 & @xmath1011 & @xmath728 & @xmath1012 & @xmath1013 & @xmath1014 & @xmath1015 & @xmath1016 & @xmath1017 + v17 & @xmath1018 & @xmath1019 & @xmath1020 & @xmath1021 & @xmath1022 & @xmath1023 & @xmath1024 & @xmath1025 + v21 & @xmath1026 & @xmath1027 & @xmath1028 & @xmath1029 & @xmath120 & @xmath1002 & @xmath468 & @xmath1030 + nv6 & @xmath1031 & @xmath1019 & @xmath1032 & @xmath1033 & @xmath1034 & @xmath1035 & @xmath1036 & @xmath1017 + nv8 & @xmath1037 & @xmath508 & @xmath1038 & @xmath1039 & @xmath1040 & @xmath1041 & @xmath1042 & @xmath1025 + nv13 & @xmath1043 & @xmath1044 & @xmath1045 & @xmath1023 & @xmath1046 & @xmath1008 & @xmath1047 & @xmath1048 + nv21 & @xmath1049 & @xmath1050 & @xmath1051 & @xmath1052 & @xmath376 & @xmath1053 & @xmath1054 & @xmath1055 + nv23 & @xmath1056 & @xmath1019 & @xmath1057 & @xmath1058 & @xmath281 & @xmath1059 & @xmath1060 & @xmath1061 + nv27 & @xmath1062 & @xmath1063 & @xmath1001 & @xmath1064 & @xmath342 & @xmath1065 & @xmath1066 & @xmath1067 + mean & @xmath1068 & @xmath1069 & @xmath1070 & @xmath1071 & @xmath1072 & @xmath1073 & @xmath1074 & @xmath1075 [ tab5 ] @xcite , @xcite , @xcite , @xcite , and @xcite derived empirical expressions that relate metallicity , absolute magnitude , and temperature with the fourier parameters of rr_ab _ stars , in the case of sufficiently `` regular '' light curves ( @xmath1076 ) .
we accordingly use equations ( 1 ) , ( 2 ) , ( 5 ) , and ( 11 ) in @xcite to compute @xmath2}$ ] , @xmath794 , @xmath1077 , and @xmath1078 , respectively , for the 12 rr_ab _ variables in our sample with @xmath1076 .
the color indices and @xmath1079 come from equations ( 6 ) and ( 9 ) of @xcite ; we then use equation ( 12 ) of @xcite , assuming a mass of @xmath1080 , to derive temperature values from equation ( 11 ) ( for ) and equation ( 12 ) ( for @xmath1079 ) in @xcite .
these results are given in table [ tab5 ] .
fourier parameters suggest a metallicity of @xmath2 } = -1.52 \pm 0.21 $ ] for ngc 5286 in the @xcite scale ; this corresponds to a value of @xmath2 } = -1.68 $ ] in the @xcite scale .
this is consistent with the value derived using the rr_c _
variables and is in excellent agreement with ( * ? ? ?
* @xmath2 } = -1.67 $ ] ) , and also with the value obtained in paper i ( @xmath2 } = -1.70\pm 0.05 $ ] , again in the @xcite scale ) on the basis of several different photometric parameters describing the shape and position of the cluster s red giant branch in the @xmath9 , diagram .
we find a mean absolute magnitude of @xmath1081 mag for the rr_ab _ stars .
for the same set of 12 rr_ab _ stars used to derive this value , we also find @xmath1082 mag .
this implies a distance modulus of @xmath1083 for ngc 5286 , which is in excellent agreement with the value provided in the @xcite catalog , namely , @xmath1084 mag . if one adopts instead for the hb an average absolute magnitude of @xmath1085 mag at the ngc 5286 metallicity , as implied by equation ( 4a ) in @xcite which is based on a calibration of the rr lyrae distance scale that uses the latest hipparcos and _ hubble space telescope _
trigonometric parallaxes for rr lyr , and takes explicitly into account the evolutionary status of this star
one finds for the cluster a distance modulus of @xmath5 mag for ngc 5286 .
we caution the reader that the ngc 5286 rr lyrae stars could in principle be somewhat overluminous for the cluster s metallicity ( in view of the cluster s predominantly blue hb ) , in which case the correct distance modulus could be even larger , by an amount that could be of the order of @xmath1086 mag ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . finally , and as also pointed out by @xcite , we also warn the reader that intrinsic colors and temperatures estimated from fourier decomposition are not particularly reliable , and should accordingly be used with due caution .
the reader is also referred to @xcite and @xcite for caveats regarding the validity of the results obtained based on the @xcite relations for rr_c _ stars .
figure [ cuantas ] shows a histogram with our derived periods in .
the bottom panel is similar to the upper panel , but with the rr_c _ periods `` fundamentalized '' using the equation @xmath1087 ( e.g. , * ? ? ?
* and references therein ) .
note the lack of a sharply peaked distribution , contrary to what is seen in several gcs of both oosterhoff types , but most notably in m3 ( e.g. , * ? ? ?
* ; * ? ? ?
* and references therein ) .
in order to assign an oosterhoff type to ngc 5286 , we must compare its rr lyrae properties with those found in other @xcite type i and ii gcs . in this sense
, @xcite found the mean rr_ab _ and rr_c _ ( plus rr_d _ ) periods for rr lyrae stars in galactic gcs to be @xmath1088 days and @xmath1089 days , respectively , for ooi clusters , and @xmath1090 days and @xmath1091 days , respectively , in the case of ooii clusters .
in addition , catelan et al .
( 2009 , in preparation ) have recently shown that the minimum period of the ab - type pulsators @xmath4 , when used in conjunction with @xmath1092 , provides a particularly reliable diagnostics of oosterhoff status . in this sense ,
the key quantities for the cluster can be summarized as follows : @xmath1093 one immediately finds that the value of @xmath1092 for the cluster points to an ooii status ( see also fig . [ pf ] , which is based on the compilation presented in @xcite @xcite ) which is also favored by its relatively high c - type number fraction . on the other hand , the average period of the rr_c
s is lower than typically found among ooii globulars , being more typical of ooi objects .
however , as shown by catelan et al .
( 2009 ) , there is a large overlap in @xmath1094 values between ooi and ooii globulars , thus making this quantity a poorer indicator of oosterhoff status than is often realized .
the situation regarding @xmath4 is rather interesting , for the value derived for ngc 5286 makes it one of the ooii clusters with the shortest @xmath4 values to date though 0.512 d would still clearly be too long for an ooi cluster , which generally have @xmath1095 d ( catelan et al . 2009 ) as opposed to ooii clusters , which typically have instead @xmath1096 d. as can be seen from table [ tab1 ] and figure [ cuantas ] ( _ top _ ) , after v7 ( the star with @xmath1097 d ) , the next shortest - period star in ngc 5286 is nv11 , with @xmath1098 d ; this is indeed less atypical for an ooii object .
the position of the cluster in a metallicity versus hb type diagram is displayed in figure [ fl ] . here
hb type @xmath1099 , where @xmath1100 are the numbers of blue , red , and variable ( rr lyrae - type ) hb stars , respectively ; this quantity was derived for ngc 5286 in paper i. as discussed by @xcite , oosterhoff - intermediate gcs , such as found in several gcs associated with the dwarf satellite galaxies of the milky way , tend to cluster inside the triangular - shaped region marked in this diagram whereas galactic gcs somehow are not found in this same region , thus giving rise to the oosterhoff dichotomy in the galaxy . in this same plane ,
ooi clusters tend to fall to the left ( i.e. , redder hb types ) of the triangular - shaped region , whereas ooii objects are more commonly found to its right .
ngc 5286 falls rather close to the oosterhoff - intermediate region in this plane , but its position is indeed still consistent with ooii status ( see also fig . 8 in @xcite @xcite ) .
figures [ baiv ] and [ baib ] show the positions of the rr lyrae stars on the bailey ( period - amplitude ) diagram for @xmath9 and @xmath552 magnitudes , respectively .
circles indicate the rr_ab _
s , whereas crosses are used for the rr_c _ s . as in figure [ hb
] , we use different symbol sizes for the variables in different radial annuli from the cluster center . also shown in this figure
are typical lines for ooi and ooii clusters , which read as follows ( see @xcite ) : @xmath1101 @xmath1102 for ab - type rr lyrae stars in ooi clusters . for rr_ab
_ s in ooii clusters , in turn , the same lines can be used , but shifted in periods by @xmath1103 . in the case of c - type stars ,
we derive reference lines on the basis of figures 2 and 4 of @xcite ; these read as follows : @xmath1104 @xmath1105 for c - type rr lyrae stars in ooi clusters , and @xmath1106 @xmath1107 for c - type rr lyrae stars in ooii clusters ( again based on presumably `` evolved '' rr lyrae stars in m3 ) . this kind of diagram can be used as a diagnostic tool to investigate the oosterhoff classification of rr_ab _ stars .
however , the position of a star in this diagram can be strongly affected by the presence of the blazhko effect . in order to minimize this problem
, we make a distinction between variables with a jurcsik - kovcs compatibility parameter value @xmath1076 ( _ filled circles _ ) and @xmath1108 ( _ open circles _ ) . even considering only the rr lyrae stars with small @xmath804 and hence presumably `` regular '' light curves ( according to the jurcsik - kovcs criterion )
, we see that there is still a wide scatter among the rr_ab _ s in the bailey diagrams , with no clear - cut tendency for stars to cluster tightly around either oosterhoff reference line , particularly in the @xmath552 case .
it is possible that at least some of the dispersion is caused by unidentified blends in the heavily crowded inner regions of the cluster , where most of the variables studied in this paper can be found ( see fig . [ ubica ] ) .
in fact , the variable stars in the innermost cluster regions ( small circles ) show more scatter .
if we only look at the variables that lie outside the core radius ( the large and medium - sized circles , respectively , with @xmath1109 ) , we can see that they cluster much more tightly around the oosterhoff ii line .
it should be noted , in any case , that very recently @xcite have shown that the ab - type rr lyrae stars in the prototypical ooii cluster m15 ( ngc 7078 ) similarly do not cluster around the ooii reference line derived on the basis of more metal - rich clusters , thus casting some doubt on the validity of these lines as indicators of oosterhoff status , at least at the more metal - poor end of the rr lyrae metallicity distribution .
as far as the positions of the c - type rr lyrae stars in the bailey diagrams are concerned , we find that there is also a wide scatter , without any clearly defined tendency for the data to clump tightly around either of the oosterhoff reference lines although the distribution does seem skewed towards shorter amplitudes at a given period , compared to the typical situation in ooii clusters .
finally , we can also check how the average fourier - based physical parameters derived for the ngc 5286 variables rank the cluster in terms of oosterhoff status .
this exercise is enabled by a comparison with the data for several clusters of different oosterhoff types , as compiled in tables 6 and 7 of @xcite .
for the rr_c _
s , we find that the mean masses , luminosities and temperatures are in fact more similar to those found for m3 ( a prototypical ooi cluster ) than they are for ooii globulars .
part of the problem may be due to the fact that ngc 5286 is significantly more metal - rich than all ooii globulars used in the analysis ; recall that @xmath1110 is the _ only _ fourier parameter used in the @xcite calibration of masses , luminosities , and temperatures , and that the impact of metallicity on the @xcite relations has still not been comprehensively addressed ( see 5 in @xcite @xcite , and also 4 in @xcite @xcite for general caveats regarding the validity of those relations ) . as a matter of fact , the recent study by @xcite clearly shows that , in the case of rr_c _
variables , @xmath1110 depends strongly on the metallicity . for the rr_ab _
s , in turn , both the derived temperatures and absolute magnitudes are fully consistent with an ooii classification for the cluster .
we find one type ii cepheid ( v8 ) with a period of @xmath89 days and a visual amplitude of @xmath1111 mag , typical for a bl herculis star .
we use equation ( 3 ) in @xcite to obtain @xmath1112 mag for v8 . using the intensity - weighted mean magnitude for v8 from table [ tab1 ]
, we obtain for the cluster a distance modulus @xmath1113 mag , slightly shorter than the values discussed in [ sec : subab ] .
however , as we can see from figure 9 in @xcite , a large dispersion in @xmath794 is indeed present for short - period type ii cepheids , thus possibly explaining the small discrepancy .
in this paper , we present the results of time - series photometry for ngc 5286 , a gc which has been tentatively associated with the canis major dwarf spheroidal galaxy .
38 new variables were discovered in the cluster , and 19 previously known ones were recovered in our study ( including one bl her star that was previously catalogued as an rr lyrae ) .
the population of variable stars consists of 52 rr lyrae ( 22 rr_c _ and 30 rr_ab _ ) , 4 lpv s , and 1 type ii cepheid . from fourier decomposition of the rr_ab _
light curves , we obtained a value for the metalicity of the cluster of @xmath2 } = -1.68 \pm 0.21 $ ] dex in the @xcite scale .
we also derive a distance modulus of @xmath5 mag for ngc 5286 , based on the recent rr lyrae @xmath1114}$ ] calibration of @xcite . using a variety of indicators ,
we discuss in detail the oosterhoff type of the cluster , concluding in favor of an ooii classification .
the cluster s fairly high metallicity places it among the most metal - rich ooii clusters known , which may help account for what appears to be a fairly unusual behavior for a cluster of this type , including relatively short values of @xmath4 and @xmath1115 , and unusual physical parameters , as derived for its c - type rr lyrae stars on the basis of fourier decomposition of their light curves . in regard to the cluster s suggested association to the canis major dwarf spheroidal galaxy
, we note that the metallicity and distance modulus derived in this work are very similar to the values previously accepted for the cluster @xcite , and thus the conclusions reached by previous authors @xcite regarding its possible association with this dwarf galaxy are not significantly affected by our new metallicity and distance estimates .
in addition , the position of the cluster in the hb morphology - metallicity plane is fairly similar to that found in several nearby extragalactic systems . as far as ngc
5286 s rr lyrae pulsation properties are concerned , the present study shows them to be somewhat unusual compared with bona - fide galactic globular clusters , but still do not classify the cluster as an oosterhoff - intermediate system , as frequently found among the galaxy s dwarf satellites ( e.g. , * ? ? ?
* and references therein ) .
it is interesting to note , in any case , that the canis major field , unlike what is found among other dwarf galaxies , appears to be basically devoid of rr lyrae stars @xcite , and also to be chiefly comprised of fairly high - metallicity ( @xmath1116 } \geq -0.7 $ ] ) , young ( @xmath1117 gyr ) stars ( e.g. , * ? ? ?
* ) . the present paper , along with paper i ,
show instead that ngc 5286 is an rr lyrae - rich , metal - poor globular cluster that is at least as old as the oldest globular clusters in the galactic halo .
it is not immediately clear that such an object as ngc 5286 would be easily formed within a galaxy with the properties observed for the canis major main body and this should be taken into account when investigating the physical origin and formation mechanism for the canis major overdensity and its associated tidal ring .
we warmly thank i. c. leo for his help producing the finding chart , and an anonymous referee for several comments that helped improve the presentation of our results .
mz and mc acknowledge financial support by proyecto fondecyt regular # 1071002 . support for mc is also provided by proyecto basal pfb-06/2007 , by fondap centro de astrofsica 15010003 , and by a john simon guggenheim memorial foundation fellowship .
has is supported by csce and nsf grant ast 0607249 .
alard , c. 2000 , , 144 , 363 bellazzini , m. , ibata , r. , monaco , l. , martin , n. , irwin , m. j. , & lewis , g. f. 2004 , mnras , 354 , 1263 bono , g. , caputo , f. , & stellingwerf , r. f. 1995 , , 99 , 263 cacciari , c. , corwin , t. m. , & carney , b. w. 2005 , , 129 , 267 caloi , v. , castellani , v. , & piccolo , f. 1987 , , 67 , 181 catelan , m. 2004a , , 600 , 409 catelan , m. 2004b , in variable stars in the local group , asp conf .
ser . , 310 , ed .
d. w. kurtz & k. r. pollard ( san francisco : asp ) , 113 catelan , m. 2007 , revmexaa conf .
ser . , 26 , 93 catelan , m. 2009 , , 320 , 261 catelan , m. , & corts , c. 2008 , , 676 , l135 catelan , m. , & de freitas pacheco , j. a. 1993 , , 106 , 1858 clement , c. m. , jankulak , m. , & simon , n. r. 1992 , , 395 , 192 clement , c. m. , muzzin , a. , dufton , q. , ponnampalam , t. , wang , j. , burford , j. , richardson , a. , rosebery , t. , rowe , j. , & hogg , h. s. 2001 , , 122 , 2587 contreras , r. , catelan , m. , smith , h. a. , pritzl , b. j. , & borissova , j. 2005 , , 623 , l117 corwin , t. m. , catelan , m. , smith , h. a. , borissova , j. , ferraro , f. r. , & raburn , w. s 2003 , , 125 , 2543 corwin , t. m. , borissova , j. , stetson , p. b. , catelan , m. , smith , h. a. , kurtev , r. , & stephens , a. w. 2008 , , 135 , 1459 crane , j. d. , majewski , s. r. , rocha - pinto , h. j. , frinchaboy , p. m. , skrutskie , m. f. , & law , d. r. 2003 , , 594 , 119 dantona , f. , & caloi , v. 2008 , , 390 , 693 demarque , p. , zinn , r. , lee , y .- w . , &
yi , s. 2000 , , 119 , 1398 forbes , d. a. , strader , j. , & brodie , j. p. 2004
, , 127 , 3394 gerashchenko , a. n. , kadla , z. i. , & malakhova , yu . n. 1997 , ibvs , 4418 , 1 harris , w. e. 1996 , , 112 , 1487 jurcsik , j. 1995 , aca , 45 , 653 jurcsik , j. 1998 , , 333 , 571 jurcsik , j. , & kovcs , g. 1996 , , 312 , 111 kinman , t. d. , saha , a. , & pier , j. r. 2004 , , 605 , l25 kovcs , g. 1998 , msait , 69 , 49 kovcs , g. , & jurcsik , j. 1996 , , 466 , l17 kovcs , g. , & jurcsik , j. 1997 , 322 , 218 kovcs , g. , & kanbur , s. m. 1998 , , 295 , 834 kovcs , g. , & walker , a. r. 1999 , , 512 , 271 kovcs , g. , & walker , a. r. 2001 , , 371 , 579 lee , j .- w . , & carney , b. w. 1999 , , 118 , 1373 liller , m. h. , & lichten , s. m. 1978 , , 83 , 41 mateu , c. , vivas , a. k. , zinn , r. , miller , l. r. , & abad , c. 2009 , , 137 , 4412 morgan , s. m. , wahl , j. n. , & wieckhorst , r. m. 2007 , , 374 , 1421 oosterhoff , p. th .
1939 , observatory , 62 , 104 oosterhoff , p. th . 1944 , bull .
, 10 , 55 pritzl , b. j. , smith , h. a. , stetson , p. b. , catelan , m. , sweigart , a. v. , layden , a. c. , & rich , r. m. 2003 , , 126 , 1381 simon , n , r. , clement , c. m. 1993 , , 410 , 526 stellingwerf , r. f. 1978 , , 224 , 953 stetson , p. b. 1987 , , 99 , 191 stetson , p. b. 1994 , , 106 , 250 zinn , r. , & west , m. j. 1984 , , 55 , 45 zorotovic , m. , catelan , m. , zoccali , m. , pritzl , b. j. , smith , h. a. , stephens , a. w. , contreras , r. , & escobar , m. e. 2009 , , 137 , 257 ( paper i ) | we present the results of a search for variable stars in the globular cluster , which has recently been suggested to be associated with the canis major dwarf spheroidal galaxy .
57 variable stars were detected , only 19 of which had previously been known . among our detections
one finds 52 rr lyrae ( 22 rr_c _ and 30 rr_ab _ ) , 4 lpv s , and 1 type ii cepheid of the bl herculis type .
periods are derived for all of the rr lyrae as well as the cepheid , and _
bv _
light curves are provided for all the variables .
the mean period of the rr_ab _
variables is @xmath0 days , and the number fraction of rr_c _ stars is @xmath1 , both consistent with an oosterhoff ii ( ooii ) type thus making ngc 5286 one of the most metal - rich ( @xmath2 } = -1.67 $ ] ; @xcite @xcite ) ooii globulars known to date . the minimum period of the rr_ab _ s , namely @xmath3 d , while still consistent with an ooii classification , falls towards the short end of the observed @xmath4 distribution for ooii globular clusters . as was recently found in the case of the prototypical ooii globular cluster m15 ( ngc 7078 ) , the distribution of stars in the bailey diagram does not strictly conform to the previously reported locus for ooii stars .
we provide fourier decomposition parameters for all of the rr lyrae stars detected in our survey , and discuss the physical parameters derived therefrom .
the values derived for the rr_c _
s are not consistent with those typically found for ooii clusters , which may be due to the cluster s relatively high metallicity the latter being confirmed by our fourier analysis of the ab - type rr lyrae light curves .
using the recent recalibration of the rr lyrae luminosity scale by @xcite , we derive for the cluster a revised distance modulus of @xmath5 mag . |
we acknowledge useful discussions with david huse and nandini trivedi .
this work used the extreme science and engineering discovery environment ( xsede ) , which is supported by national science foundation grant number oci-1053575 and the high performance computing cluster at case western reserve university .
22ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevb.63.014413 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.72.104404 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.98.117205 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.77.224413 [ * * , ( ) ] link:\doibase 10.1126/science.1201080 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop `` ''
http://link.aps.org/doi/10.1103/physrevb.45.12377 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.95.176406 [ * * , ( ) ]
link:\doibase 10.1103/physrevb.83.100404 [ * * , ( ) ] @noop * * , ( ) @noop `` , '' @noop * * , ( ) link:\doibase 10.1103/physrevlett.107.087204 [ * * , ( ) ] @noop * * , ( ) http://link.aps.org/doi/10.1103/physrevb.65.165113 [ * * ( ) ] @noop @noop ( , ) @noop * * , ( ) @noop * * , ( ) | the dirac spin liquid ground state of the spin 1/2 heisenberg kagome antiferromagnet has potential instabilities @xcite .
this has been suggested as the reason why it is not strongly supported in large - scale numerical calculations@xcite .
however , previous attempts to observe these instabilities have failed .
we report on the discovery of a projected bcs state with lower energy than the projected dirac spin liquid state which provides new insight into the stability of the ground state of the kagome antiferromagnet .
the new state has three remarkable features .
first , it breaks both spatial symmetry in an unusual way that may leave spinons deconfined along one direction .
second , it breaks the u(1 ) gauge symmetry down to @xmath0 .
third , it has the spatial symmetry of a previously proposed `` monopole '' suggesting that it is an instability of the dirac spin liquid .
the state described herein also shares a remarkable similarity to the distortion of the kagome lattice observed at low zn concentrations in zn - paratacamite suggesting it may already be realized in these materials .
many potential ground states have been suggested for the spin 1/2 kagome heisenberg antiferromagnet ; these include magnetic ordering in one of several spin patterns@xcite , a valence bond crystal with a 36 site unit cell@xcite or a 12 site unit cell@xcite , a chiral spin liquid@xcite , several kinds of @xmath0 spin liquids@xcite and an algebraic spin liquid@xcite . amongst algebraic spin liquids ,
it was shown @xcite that the @xmath1-dirac state had lowest variational energy .
this latter state is characterized by its graphene - like dirac fermionic spinons interacting with a maxwell - like photon that characterizes singlet excitations .
many of the other suggestions for the kagome antiferromagnet ground state can be characterized as instabilities of this dirac spin liquid ; these instabilities have been catalogued in ref . .
however , despite the many instabilities , all variational studies we are aware of that include the @xmath1-dirac state for some choice of variational parameters have found it remarkably stable@xcite .
recently , large scale dmrg calculations@xcite have produced strong numerical evidence that the true ground state of the kagome antiferromagnet is a @xmath0 spin liquid .
in particular , they have found the lowest known variational state with energies comparable to exact diagonalization on small clusters , a gap to all excitations and topological entanglement entropy expected of a @xmath0 spin liquid .
these results are quite surprising in light of prior studies of the u(1)-dirac state and suggest that it is unstable .
perhaps the simplest explanation of the dmrg results , as proposed in ref .
, is that the dirac - like spinons pair up and form a superfluid - like @xmath0 spin liquid . with this goal in mind ,
these authors catalog 20 such possible states and give explicit instructions for constructing 14 of them . unfortunately , despite searches for such a state@xcite , no @xmath0 state has been shown to be energetically favorable to the @xmath1 dirac state at the level of variational projected bcs wave functions .
correlations for the projected dirac spin liquid state with twisted boundary conditions ( left ) and the lowest energy projected mean field state ( right ) .
the color indicates sign and the linewidth measures the magnitude . , scaledwidth=40.0% ] in this letter , we revisit the projected bcs ( pbcs ) variational wave function problem on the kagome lattice and have discovered an instability of the dirac state .
our approach , building on refs . , is to optimize over the entire set of time - reversal invariant projected bcs wave functions which are parameterized by a symmetric pairing matrix @xmath2 .
these states encompass the u(1 ) and z2 spin liquids , general valence bond solids , and a number of other instabilities .
we find states with lower energy than the dirac state on clusters of up to 192 sites . on the 48 site cluster in particular , a carefully optimized wave function , as shown in fig .
[ fig : s.s ] , demonstrates that the new state breaks lattice symmetries , doubling the unit cell with a wave vector that connects the two dirac nodes .
however it preserves a @xmath3 rotational symmetry about several lattice points giving it a distinct one dimensional character .
it also lies very close to the dirac state with similar fluxes through the hexagons and bow ties . for these reasons ,
we believe it should belong to one of the leading instabilities proposed in ref .
, and have discovered that the most likely candidate is the time - reversal symmetric `` @xmath4-monopole '' .
we also find that this state can not be obtained by optimizing within the class of wave functions that do not admit pairing and therefore likely has @xmath0 global gauge symmetry .
for this reason and because of its @xmath3 symmetry and one dimensional character , we have dubbed this state a @xmath0 striped ( or smectic ) spin liquid crystal .
finally , we discuss a connection between our state and zn - paratacamite with zn concentrations @xmath5 below @xmath6 whose observed distorted kagome layers@xcite has the same symmetry as fig .
[ fig : s.s](b ) .
our objective is to find the ground state of the nearest neighbor spin @xmath7 kagome antiferromagnet constrained to the set of gutzwiller projected variational bcs wave functions @xmath8
where @xmath9 , @xmath10 is the bcs pairing amplitude and @xmath11 projects these states onto the physical singly occupied hilbert space of spin wave functions .
this set of variational ansatz includes projected slater determinants such as the @xmath1 dirac spin liquid @xcite .
the variational search is performed by minimizing the ground state energy @xmath12 of the heisenberg spin hamiltonian on the kagome lattice using as parameters all @xmath13 real values of the pairing function @xmath14 ( building on refs . ) . to avoid only finding local minima in this energy landscape
, we choose to start the optimization in many qualitatively different starting points by making use of the well developed projective symmetry group classification developed in the literature@xcite .
in particular , we will choose mean field parameters following ref . and derive a pairing matrix from these following supplemental materials(sm ) s - i and s - ii .
this allows us to start from 14 different spin liquid wave functions .
fig .
[ fig : opt](a ) shows prototypical optimization traces of the energies for each spin liquid state on the @xmath15 lattice , labeled following table ii of ref . .
many states lie below the energy of the dirac spin liquid and the minimal energy state we find is at @xmath16 per site ( see sm s - iii for its pairing matrix ) , significantly below the dirac spin liquid energy of @xmath17 per site showing that the dirac spin liquid is not the most stable projected mean field state .
this is particularly surprising in light of previous studies that find the dirac spin liquid to be stable against many instabilities@xcite .
such calculations failed to find this state because they only considered short - range mean field hamiltonians , a limitation we avoid by optimizing a general pairing function that allows both long - range hopping terms and spatial symmetry breaking in the fermion hamiltonian .
as shown in fig .
[ fig : opt](b ) and discussed in more detail below , our approach in conjunction with enforcing @xmath1 symmetry does not lower the energy .
but , additionally allowing the @xmath1 gauge symmetry to break down to @xmath0 does lower the energy ( left side of fig . [ fig : opt ] ) .
consequently , the gauge symmetry breaking down to @xmath0 is intricately linked to spatial symmetry breaking and also possibly long - range hopping / pairing terms in the hamiltonian .
let us then focus on the nature of our lowest energy state .
most strikingly we find that the state breaks translational symmetry doubling the unit cell .
fig .
[ fig : s.s ] ( right ) shows the deviation of the spin - spin correlation function on nearest neighbors from its average value , in contrast with the u(1 ) dirac spin liquid shown in fig . [
fig : s.s ] ( left ) .
the symmetry breaking seen in the dirac spin liquid is a finite size artifact coming from twisted boundary conditions in one direction . in our optimized state
we find breaking of the translational symmetry and a doubling of the unit cell leading us to conclude that the minimal projected mean field state is _ not _ an isotropic spin liquid . to further verify that there is not a nearby isotropic spin liquid state to our optimal state
, we stochastically modify the pairing function , starting from a low energy state , in small increments searching for a true symmetric spin liquid .
in particular , we make a random stochastic change to the pairing function accepting it only if we have moved closer to a spin liquid state ( defined by the deviations of the unprojected @xmath18 ) . in this way
, we search for the `` closest '' spin liquid . when we run this procedure , the spin liquid state it approaches has the energy of the dirac spin liquid ( and hence is presumably the dirac spin liquid ) .
this leads us to believe that their is no additional nearby isotropic spin liquid .
having found a broken symmetry state , we would like to now understand whether it retains characteristics reminiscent of any spin liquid phase . to make such a connection , it is useful to compute the anomalous density matrix from the unprojected wave function defined by the pairing matrix that corresponds to our lowest energy projected state
. it is given by @xmath19 where @xmath20 and transforms under an su(2 ) gauge transformation , @xmath21 , like @xmath22 exactly like the mean fields in the corresponding slave particle theory ( see ref .
@xcite or the sm s - i ) .
we can therefore use this matrix to study the projective symmetry properties of the obtained unprojected state @xmath23 . to study fluxes through the lattice
, we can use the su(2 ) matrix version of a `` phase '' variable @xmath24 that is an analog of the @xmath1 phase @xmath25 we associate with ordinary electricity and magnetism on a lattice . following ref .
, the analog of flux through any loop on the lattice is therefore the product of this matrix around the loop @xmath26 where @xmath27 is the number of bonds @xmath28 that form the loop .
unfortunately , this product is not gauge invariant , but for every loop of the lattice , we can define an angle @xmath29 associated with its flux through the trace of this matrix . unlike a @xmath1 flux ,
here @xmath30 introduces a phase change of @xmath31 .
the natural first loops to consider when characterizing the state @xmath32 are the nearest neighbor bow ties ( product of two neighboring triangles ) and hexagons ( the trace of @xmath33 for an odd site loop vanishes by time reversal symmetry ) .
these loops allow us to determine which of the four u(1 ) spin liquid states is closest to our best optimized state .
the results are @xmath34 where the error estimate is the standard deviation and @xmath35 denotes the average value of the flux over different hexagons or bow ties respectively .
these results are equivalent to nearly @xmath36 flux through the hexagon and @xmath37 flux through the bow tie or triangle in the traditional @xmath1 description of flux .
they show distinctly that the optimized state is very close to the @xmath1 dirac state as expected from energetic considerations but that because of the symmetry breaking shown in fig
. [ fig : s.s ] it is an instability . having established that our newly discovered state is very close to the dirac state , we now turn to its symmetry breaking properties . to this end , we seek the space group representations that give the largest contribution to the pattern in fig . [
fig : s.s](b ) which are periodic with at most a quadrupled unit cell .
such representations were constructed in ref .
@xcite and labeled as @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 for those related to the point group alone and @xmath44 , @xmath45=@xmath46 , @xmath47 , @xmath48 for those allowed by a doubling / quadrupling of the unit cell . the focus of ref .
@xcite was on the @xmath44 representation for the `` hastings valence bond crystal '' states associated with the generation of mass of the dirac fermions .
however , the bond amplitudes plotted in fig .
[ fig : s.s](b ) are not of this representation .
instead , they are dominated by the @xmath45 and @xmath43 representations whose patterns are shown in fig .
[ fig : symmetryanalysis ] ( a ) and ( b ) .
remarkably , the symmetry of the @xmath45 pattern alone is the same as the symmetry of the @xmath45 and @xmath43 patterns . the @xmath43 pattern alone , however , has higher symmetry .
hence , the symmetry breaking observed here arises uniquely from a desire to form the @xmath45 pattern .
the only time reversal symmetric alternative to the hastings states , among instabilities of the dirac fixed point identified in ref .
@xcite is the spin singlet / nodal triplet `` @xmath4-monopole '' that is created by a complex operator @xmath49 , @xmath50@xcite . in sm s - iv
we show , following the transformation properties determined in ref .
@xcite , that the six dimensional vector @xmath51 transforms under the two three - dimensional representations @xmath44 and @xmath45 .
this remarkable coincidence allows us to conjecture that the @xmath4-monopole is responsible for the instability of the dirac state observed in fig .
[ fig : s.s](b ) . to further understand the role of the symmetry breaking
, we measure the asymmetry of fig . [
fig : s.s ] during part of an optimization run and correlate this with the observed energy .
we compare global asymmetry ( removing the @xmath43 representation to remove the effects of twisted boundary conditions ) defined as @xmath52 as well as the assymetry of the @xmath45 component defined as @xmath53 where @xmath54 on the solid thick bonds , @xmath55 on the dashed thick bonds and zero otherwise with @xmath56 the number of nearest neighbor bonds .
these are shown in fig .
[ fig : symmetryanalysis ] ( c ) and ( d ) .
interestingly , although we see a clear symmetry breaking , lower energy states have lower asymmetry , saturating at a value ( triangle ) that is small but still above the wen dirac state ( star ) .
we find exactly the opposite behavior for the overlap with the @xmath45 component : lower energy states have greater overlap .
this implies the symmetry breaking associated with the @xmath45 representation is an important feature of our state throughout much of the minimization process .
in addition to breaking the spatial symmetries of the lattice , the physical system also breaks the @xmath1 symmetry of the dirac state down to a @xmath0 symmetry . to study this symmetry breaking we produced a series of runs without any pairing in the wave functions .
we initialize twelve runs by taking the dirac pairing function and multiplying each element by @xmath57 where @xmath58 is sampled in the interval @xmath59 $ ] . using @xmath60 we ran 12 simulations
whose variation in initial energy was significant starting as high as @xmath61 per site .
all simulations converged to @xmath62 , the energy of the dirac state . from these results
we conclude that the dirac state lies at the bottom of a deep and wide region in orbital space and that the pairing of spinons is necessary to produce the state shown in fig .
[ fig : s.s](b ) . based on our results
, it seems very likely that the dirac fixed point is unstable to the formation of a stripe - like spin - liquid crystal phase .
this conclusion rests on two assumptions : that fluctuations beyond those captured by the projected wave function do not restore the symmetry and that this symmetry breaking is not a finite size effect .
we have some indirect evidence supporting the latter assumption .
dmrg and exact diagonalization results indicate that a 4x4 unit cell cluster should be large enough to capture the qualitative physics of the system .
in addition , we have looked at up to 8x8 unit cell clusters and can still find states with lower energy than the dirac state .
one way to directly address these assumptions would be to perform a psg analysis on the relevant lower symmetry subgroup of the kagome lattice and use it to search for the state that projects to our state ( we can not do this directly because projection is a many - to - one mapping and can not be inverted ) .
such an analysis would allow a determination of the finite size scaling of the symmetry breaking effects and provide a starting point for studying fluctuations about this phase in a low energy effective theory .
this would also establish more directly the question of whether there is an energy gap ( however , we expect such a gap because the wave vector of the symmetry breaking pattern connects the dirac nodes of the brillouin zone ) .
given the lack of evidence in large - scale dmrg calculations for our state , it is unlikely to represent the true ground state .
however , since it lies very close by and involves a minimal loss of crystal symmetries , it seems likely to be a leading instability . in particular , small perturbations to the hamiltonian could stabilize it suggesting it could be realized in nature . one promising class of materials are the zn - paratacamite family parameterized by zn doping concentration @xmath5 with @xmath63 . unlike the structurally perfect kagome lattice of the @xmath64 herbertsmithite member of the family ,
compounds with @xmath63 , including clinoatacamite at @xmath65 , break crystal symmetries and have distorted kagome layers@xcite with precisely the distortion expected from the symmetry breaking of our state .
our results therefore motivate the study of single crystals of these materials and suggests that either the mysterious intermediate phase below @xmath66 or the high temperature phase @xmath67 could still have spinons as low energy excitations that are delocalized along the `` rails '' of the distorted lattice .
the most remarkable implication of our results is its suggestion that spin liquid crystal phases may be a common phenomena .
since any dimer state is the exact ground state at the mean field level@xcite , projection must introduce quantum fluctuations that melt such crystalline phases , take the system through a succession of more symmetric phases until , in our case , it nearly reaches an isotropic phase .
such a picture has several implications for the dmrg calculations on the kagome lattice .
it is known@xcite that small perturbations to the hamiltonian in dmrg ( boundary conditions , pinning fields , etc . )
can enhance different states and lead to symmetry breaking .
exploring the class of perturbations that stabilize the symmetry breaking we observe here would help make a deeper connection between analytic schwinger - fermion theory and dmrg .
more interestingly , the observed dmrg state might be understood as a further instability of our state to a nematic spin liquid crystal , a state found recently on a triangular lattice model with ring exchange@xcite .
a nematic spin liquid phase would be indistinguishable from a spin liquid phase on long cylinders that explicitly break rotational symmetry .
of course , it is also possible that such a putative nematic state melts into an isotropic @xmath0 spin liquid
. more generically , our results suggest that spin liquid crystal states are common in frustrated antiferromagnets and may be found using methods similar to ours as competitive ground states in many other systems . |
it is well known that in quantum chromodynamics ( qcd ) at temperature @xmath9 and quark chemical potential @xmath10 , i.e. in vacuum , quarks are in chiral symmetry breaking phase owing to the quark - antiquark condensates @xmath11 @xcite .
recently it has also been clear that , at low temperature and high baryonic density , the quarks will be in color superconducting phase owing to some diquark condensates being non - zeroes @xcite .
research on diquark condensates and color superconducting has also been extended to the region of low temperature and moderate quark chemical potential where perturbative or present - day lattice calculations are inaccessible , by means of different phenomenological models including the nambu - jona - lasinio ( njl)-type models @xcite .
since the interactions between two quarks in the antitriplet of the color @xmath12 gauge group are attractive , one can reasonably inquire if the diquark condensates could be generated in vacuum .
this problem has been researched or touched on by means of some phenomenological models @xcite . some results from a few given njl - type models
were obtained .
however , it seems that , for a thorough understanding of the diquark condensates in a njl model in vacuum , especially of the interplay between the diquark and the quark - antiquark condensates in vacuum , a general and systematic analysis is still necessary .
in addition , such research itself has certainly theoretical interest if one treats the njl model as a useful field theory model .
the njl model as a phenomenological description of qcd , is usually used in discussion of chiral symmetry breaking but very little in the diquark condensation problem in vacuum .
then what is the theoretical reason of this situation ?
in fact , in any njl model , owing to the fierz transformation @xcite , one is always allowed to include both the interactions of @xmath13-form and @xmath14-form which could lead to generation of the quark - antiquark and the diquark condensates respectively . as long as applying a njl model to diquark condensation problem in vacuum
, one will inevitably face to mutual competition between the diquark condensates and the quark - antiquark condensates .
a serious research on such mutual competition in a njl model will lead to some condition in which the diquark condensates could be generated , or say , removed in vacuum . by this condition , combined with the phenomenological requirements of qcd in vacuum ,
one will be able to answer why in conventional njl model description of qcd in vacuum one can completely neglect existence of the diquark condensates .
in fact , the condition to remove the diquark condensates in vacuum will become a useful restriction to any `` realistic '' njl model . + motivated by the above ideas , in this paper we will make a general analysis of the mutual competition between the quark - antiquark and diquark condensates in a njl - type model in vacuum .
the paper is arranged as follows . in sect.[model ] we will present the model and derive its effective potential , in sect.[ground ] determine the ground states in different conditions and in sect.[concl ] come to our conclusions and discussions .
we start with a two - flavor njl model which could be used to simulate qcd and derive its effective potential containing the order parameters respectively corresponding to quark - antiquark and diquark condensates in mean field approximation .
as is well known , in the mean field approximation , there is so called fierz ambiguity @xcite . however , throughout the whole discussions in this paper , we will follow ref.@xcite and use the fierz - transformed four - fermion couplings in the hartree approximation to avoid double counting . to simplify the problem and to make it possible to conduct the demonstration in a completely analytic way , we will first omit the small bare masses of the @xmath15 and @xmath16 quarks ( i.e. take the chiral limit ) .
thus the lagrangian of the two - flavor njl model may be written in the well - known form by @xmath17\nonumber\\ & & + h_s\sum_{a=2,5,7}(\bar{q}i\gamma_5\tau_2\lambda_aq^c ) ( { \bar{q}}^ci\gamma_5\tau_2\lambda_aq),\end{aligned}\]]with the quark fields @xmath18 in the @xmath19 doublet and the @xmath12 triplets , i.e. @xmath20 where the subscripts @xmath21 denote the three colors ( red , green and blue ) of quarks . in eq.(1 ) we have used the denotations @xmath22 , pauli matrices @xmath23 acting in the two - flavor space , the gell - mann matrices @xmath24 acting in the three - color space and the charge conjugations @xmath25 and @xmath26 of the quark fields .
the lagrangian ( 1 ) is @xmath27-invariant . for the sake of making a general discussion of mutual competition between the diquark and the quark - antiquark condensates in the model
, we temporarily view the positive coupling constants @xmath3 and @xmath4 as independent and changeable parameters .
assume that the four - fermion interactions can lead to the scalar quark - antiquark condensates @xmath28which will break the chiral symmetries @xmath29 down to @xmath30 and induce three massless pions as corresponding goldstone bosons .
we also suppose that the four - fermion interactions could generate the scalar color - antitriplet diquark and color - triplet di - antiquark condensates ( after a global @xmath12 transformation ) @xmath31it is indicated that in view of the structure of the matrix @xmath32 in @xmath33 and @xmath34 in eq.(3 ) , only the red and green quarks participate in the diquark condensates and the blue quarks do not enter them @xcite . for acquiring a qualitative understanding of the diquark condensates in vacuum , we can assume ,
similar to the bcs ansatz in superconducting theory , physical vacuum as a coherent state of red and green up and down quarks with zero total momentum @xcite which contains both quark - quark paring and antiquark - antiquark paring .
it is indicated that the njl model in vacuum i.e. at @xmath35 is a relativistic quantum field theory . hence , despite of absence of net quarks in vacuum , in such theory it is possible that the quark - quark condensates and the antiquark - antiquark condensates are generated simultaneously in vacuum , such as shown by @xmath33 and @xmath34 in eq.(3 ) .
this case is also similar to generation of quark - antiquark condensates in vacuum , where physical vacuum can be assumed to be a coherent state composed of quark - antiquark pairs with zero total momentum@xcite . as in the two - flavor color superconducting theory
, the condensates ( 3 ) will leave the chiral symmetries @xmath36 , a `` rotated '' electric charge @xmath37 and a `` rotated '' quark number @xmath38 symmetry unbroken , but will break the @xmath12 symmetry down to a @xmath39 symmetry @xcite . as a result , if the @xmath12 group is gauged , then five of the eight gluons will receive a mass through the higgs mechanism , and if the @xmath12 group is global , then five massless diquark excitations , which could be some combinations of quark - quark pairs and antiquark - antiquark pairs , will arise as the goldstone bosons of spontaneous breaking of @xmath12 .
we note that the flavor chiral symmetry breaking induced by the quark - antiquark condensates and the color symmetry breaking induced by the diquark condensates are independent each other , so even if the two condensates coexist in ground state , the total physical effects will be only a simple adding up of the effects from the two - forms of symmetry breaking .
however , in this paper , we will omit further discussions of the possible physical consequences of the diquark condensates in vacuum , and pay our main attention to the conditions to generate the two - forms of condensates and mutual competition between them .
now define that @xmath40then in mean - field ( hartree ) approximation @xcite and in the nambu - gorkov ( ng ) basis @xcite with the denotations @xmath41 the lagrangian ( 1 ) may be rewritten by @xmath42 in momentum space , the inverse propagator @xmath43 has the expression @xmath44different from the thermodynamic potential at finite temperature and finite chemical potential case , the effective potential in vacuum corresponding to the lagrangian ( 5 ) will be relativistic - invariant
. it can be obtained by a direct generalization of the conventional one order - parameter effective potential formula @xcite to the case with two order - parameters and in the ng basis .
it can be expressed by @xmath45where @xmath46 represents the propagator for massless quarks in the ng basis and the tr is to be taken over flavor , color , nambu - gorkov and dirac spin degrees of freedom .
the factor @xmath47 is due to use of the ng basis .
since the blue quarks do not enter the diquark condensates , we will have @xmath48where subscripts @xmath49 and @xmath50 mean that the corresponding matrices are limited to ( red , green ) and blue degrees of freedom in color space respectively . by using the mathematical formula @xmath51,\ ] ] where in the above @xmath52 matrix , @xmath53 , @xmath54 , @xmath55 and @xmath56 can be operators ( matrices ) , we obtain @xmath57 where @xmath58 but it is now acting in the @xmath49 two - color space and @xmath59 represents the @xmath52 unit matrix in the two - flavor space . substituting eqs.(9 ) and ( 10 ) into eq.(8 ) , we may transform eq.(7 ) to @xmath60by using the formula @xmath61 it is easy to complete the integrations and obtain the final explicit expression of the effective potential @xmath62,\ ] ] where we have made the wick rotation of the integration variable @xmath63 and introduced the 4d euclidean squared momentum cutoff @xmath64 which is assumed to satisfy the conditions @xmath65 and @xmath66 .
it should be indicated that because the 4d euclidean squared momentum cutoff is used , the effective potential expressed by eq.(12 ) will be lorentz - invariant and this feature is essential for description of the diquark condensates in vacuum . as a comparison , we note that a similar expression to eq.(12 ) can be obtained by taking the limit @xmath35 in some corresponding thermodynamical potential but a 3d momentum cutoff is usually used there .
such a limit expression of the thermodynamical potential , rigorously speaking , is not suitable to the problem of the diquark condensates in vacuum .
since the 3d momentum cutoff breaks the lorentz invariance , the relevant discussions will be theoretically inconsistent to generating mechanism of the diquark condensates in vacuum which must be built on a relativistic basis , as stated above .
hence only the effective potential ( 12 ) with the 4d momentum cutoff will give a rigorous and consistent relativistic description of the problem involving the diquark condensates in vacuum .
the relativistic effective potential given by eq.(12 ) contains two order parameters @xmath67 and @xmath68 .
a obvious advantage of the expression ( 12 ) is that one could analytically find out the minimums of @xmath69 and then determine the ground state of the system .
simultaneously , one will be able to examine the mutual competition between @xmath67 and @xmath68 when the values of the coupling constants @xmath3 and @xmath4 are changed .
the extreme value conditions @xmath70and @xmath71it is easy to see that eqs.(13 ) and ( 14 ) may have nonzero @xmath67 and @xmath68 solutions if @xmath72and @xmath73the two conditions could be satisfied when either the momentum cutoff @xmath74 is large enough for given coupling constants @xmath3 and @xmath4 or @xmath3 and @xmath4 are large enough for a given @xmath74 .
from the second order derivatives of @xmath75 denoted by @xmath76 we may define the expression @xmath77the equations ( 13 ) and ( 14 ) have four different solutions which will be discussed successively as follows .
+ 1 ) @xmath78 .
it is a maximum point of @xmath75 , since in this case we have @xmath79 @xmath80 if considering the necessary conditions ( 15 ) and ( 16 ) for existence of non - zero @xmath67 and @xmath68 solutions .
+ 2 ) @xmath81 , where @xmath82 and obeys the equation @xmath83it may be obtained by means of eq.(19 ) that when @xmath74 is large , we have @xmath84/\pi^2>0\ ] ] and @xmath85 they indicate that @xmath86 will be a minimum point of @xmath75 when @xmath5 , however , it will not be an extreme value point of @xmath75 when @xmath87 . + 3 ) @xmath88 , where @xmath89 and obeys the equation @xmath90 in this case we obtain that @xmath91 @xmath92/\pi^2.\ ] ] obviously , the sign of @xmath93 depends on @xmath53 s one .
we will have @xmath94 , if @xmath95we note that @xmath96owing to eq.(16 ) .
equation(21 ) certainly includes the case of @xmath97 .
thus @xmath88 will be a minimum point of @xmath75 only if eq.(21 ) is satisfied , i.e. either the scalar quark - antiquark interactions do not exist or they are very weak , or in other words , the scalar diquark interactions are strong enough opposite to the scalar quark - antiquark interactions .
+ 4)@xmath98 .
the none - zero @xmath99 and @xmath100 obey the equations @xmath101@xmath102now from the calculated results by using eqs.(23 ) and ( 24 ) that @xmath103>0,\ ] ] @xmath104 we may deduce that so long as a non - zero solution @xmath105 exists , it will be a minimum point of @xmath75 .
the remaining problem is to find out the condition in which @xmath99 and @xmath100 obeying eqs.(23 ) and ( 24 ) could simultaneously be equal to non - zero . for this purpose , we note the fact that the function @xmath106 has the feature that @xmath107 i.e. @xmath108 will be a monotonically increasing function of @xmath109 for large enough @xmath64 . hence , if @xmath110 , we will have the inequality @xmath111.\]]applying eq.(25 ) to eqs.(23 ) and ( 24 ) , we are led to the condition @xmath112on the other hand , by eqs.(23 ) and ( 24 ) we have the inequality @xmath113 which may be changed into @xmath114hence only if eqs.(26 ) and ( 27 ) are satisfied , we can have a non - zero solution @xmath115 as a minimum point of @xmath75 .
it may be seen by comparing eq.(21 ) with eq.(27 ) that the minimum points @xmath116 and @xmath117 can not coexist , so in each case the only minimum point ( for @xmath118 ) will correspond to the ground state of the system , especially the solution @xmath115 implies coexistence of the quark - antiquark and diquark condensates in the ground state . + all the above results can be summarized as follows .
locations of the minimum points of the effective potential @xmath75 depend on relative values of @xmath3 and @xmath4 .
the minimum points will respectively be at @xmath119 where @xmath120 and @xmath121 are determined by eqs.(19 ) and ( 20 ) separately , and @xmath105 by eqs.(23 ) and ( 24 ) . combining eq.(28 ) with eqs.(15 ) and ( 16 ) , we have shown in fig.1 the corresponding regions to the three different phases in the coupling constants @xmath2 plane .
+ ( location of fig.1 ) it may be seen that in the region with non - zero condensates , the pure quark - antiquark condensate phase and the pure diquark condensate phase is separated by a coexistence phase of the two condensates . in each phase , either ( @xmath122,@xmath121 ) or ( @xmath99,@xmath100 ) or ( @xmath120,0 ) is the only minimum point of @xmath75 and no other minimum point arises , so there is no metastable state in the model .
we indicate that the above @xmath2 phase diagram in the vacuum of a njl model is first obtained .
+ the results show that the pure quark - antiquark condensates phase and the phase with diquark condensates is distinguished merely by the ratio @xmath123 and the diquark condensates could emerge only if @xmath124 and @xmath6 is below the critical value 2/3 . from the expressions ( 11 ) and ( 12 ) of the effective potential @xmath75 and the follow - up discussions it is not difficult to see that the critical value 2/3 reflects the following fact : in the two - flavor njl model only the two colors ( red and green ) of quarks anticipate in the diquark condensates but all the three colors ( red , green and blue ) of quarks get into the quark - antiquark condensates .
we have analyzed the effective potential of a two - flavor njl model at @xmath35 which contains two order parameters @xmath67 and @xmath68 separately corresponding to quark - antiquark and diquark condensates and proven that the mutual competition between @xmath67 and @xmath68 is decided by the ratio of the scalar quark - antiquark and the scalar diquark four - fermion coupling constant @xmath3 and @xmath4 .
the results indicate that even in vacuum , the diquark condensates could either exist alone or coexist with quark - antiquark condensates when @xmath125 and @xmath4 is large enough .
however , only the pure quark - antiquark condensates could be generated when @xmath5 .
the last conclusion coincides with the result obtained in ref .
@xcite based on a random matrix model , where the critical value of @xmath6 above which no diquark condensates arise is @xmath126 for a color @xmath127 group . in fact , generalization of the analysis in present paper to @xmath128 case is direct . in doing so ,
not only the critical value @xmath126 will be derived but also a complete @xmath2 phase diagram similar to fig.1 can be obtained . the conclusion that a very large coupling constant @xmath4 in an njl model could lead to diquark condensates in vacuum
is also touched on in ref.@xcite , but no further details were given there .
it is also indicated that , in the above both models no corresponding @xmath2 phase diagram was given .
+ the above conclusions are reached in the chiral limit .
if a non - zero degenerate bare quark mass @xmath129 is included , then the effective potential will have a little more complicated expression thus a complete analytic demonstration will become more difficult .
however , the essential part of the discussions including determination of the conditions of emergence of the minimum points @xmath86 and @xmath105 can still be conducted analytically and the relevant essential conclusions reached in the @xmath130 case will keep unchanged . in particular , we can definitely give the critical value of @xmath6 at which the pure quark - antiquark condensate phase will transfer to the coexistence phase of quark - antiquark and diquark condensates , but now the value will change from 2/3 into @xmath131 , where @xmath132 is the value of the order - parameter related to the quark - antiquark condensates in the ground state . because the extra factor @xmath133 , so in nor - zero bare quark mass case , appearance of the diquark condensates requires stronger diquark interactions than the ones in the chiral limit . + in our discussions , the coupling constants @xmath3 and @xmath4 have been viewing as independent and changeable parameters , this is only a theoretical assumption for the purpose of exclusively researching mutual competition between the diquark and the quark - antiquark condensates in the model .
in fact , @xmath3 and @xmath4 are interrelated via the fierz transformations . therefore , for a given njl model in advance , the ratio @xmath6 in eq.(28 ) must be understood as its derived value in the hartree approximation after the fierz transformations are implemented .
+ by the above results , it seems that , theoretically , a general two - flavor njl model does not remove existence of diquark condensates in vacuum , as long as the diquark interactions are strong enough .
however , this can not be realistic case .
since a njl model is usually used as a low energy phenomenological description of qcd , its coupling forms and strengths must be restricted by the underlying qcd theory and/or phenomenology , hence the ratio @xmath6 can not be arbitrary .
if the underlying lagrangian or all the couplings of @xmath13-form are known , then via the fierz transformations , the effective coupling @xmath3 including direct and exchange interactions and the coupling @xmath4 in the diquark channel can be fixed uniquely in the hartree approximation , and so is the ratio @xmath6 . however ,
when the underlying lagrangian is unknown , one can write down only the partial coupling terms of @xmath13-form by symmetries and/or phenomenology , then the effective @xmath3 and @xmath4 derived by the fierz transformations ( still in the hartree approximation ) will not be unique , since they will change with different selections of the undetermined coupling terms of @xmath13-form .
+ however , our theoretical results ( 28 ) , combined with phenomenological fact that there is no diquark condensates in the vacuum of qcd , will be able to place a useful restriction to the above different selections .
in fact , for any njl model which is intended to simulate qcd , whatever initial coupling terms as a starting point of the model building are selected , via the fierz transformations , the derived least value of the effective ratio @xmath6 of the positive scalar quark - antiquark and the positive scalar diquark coupling constants in the hartree approximation must be larger than 2/3 .
this restriction is applicable to any realistic njl model . +
two examples of the models which obey the above restriction are the models where the four - fermion interactions are assumedly induced by heavy gluon exchange or by instantons in qcd .
these interactions can considered as some underlying more microscopic ones . in both models ,
the resulting ratio @xmath6 via the fierz transformations are equal to 4/3 @xcite which is obviously larger than the critical value 2/3 for generation of the diquark condensates .
so in these models we will have only the quark - antiquark condensates surviving . in ref.@xcite
a slightly different conclusion was reached in an instanton - induced model with fixed values of @xmath3 and @xmath4 . there for @xmath134 case , in the ground state similarly only the quark - antiquark condensates exist , but a pure diquark condensates also exist in a metastable state .
however , it is noted that in that model the corresponding ratio @xmath6 was not taken to be the conventional value 4/3 and a further relation between the coupling constants and the instanton density was used . +
another more phenomenological model is the conventional chiral @xmath36-invariant four - fermion model with interactions @xcite @xmath135.\ ] ] via the fierz transformations we may obtain the effective lagrangian @xmath136 with @xmath137 where the ellipses indicate the possible vector , axial - vector and tensor coupling terms .
it is emphasized that the effective lagrangian @xmath138 is used only in the hartree approximation @xcite .
for the two - flavor and three - color case , we get that the corresponding effective coupling constants @xmath139 obviously , owing to that @xmath140 , it is impossible to generate the scalar diquark condensates in this model .
the maximal attractive channels are still the original starting point , i.e. the coupling terms @xmath13 and ( @xmath141 . the possible condensates will be @xmath0 ( and/or the @xmath142 condensates @xmath143 ) and they are related to spontaneous chiral symmetry breaking only .
+ in brief , in a few known possibly realistic qcd - analogous two - flavor njl models , owing to that the condition to generate the diquark condensates in vacuum is not satisfied , there could be only the quark - antiquark condensates in ground states of these models at @xmath35 .
these models are merely suitable for description of chiral symmetry breaking and have nothing to do with color superconductivity at @xmath35 .
this certainly reflects the reality of qcd in vacuum .
+ 99 y. nambu , phys .
lett . * 4 * ( 1960 ) 380 . y. nambu and g. jona - lasinio , phys . rev .
* 122 * ( 1961 ) 345 .
j. goldstone , nuovo cimento * 19 * ( 1961 ) 154 . y. nambu and g. jona - lasinio , phys . rev .
* 124 * ( 1961 ) 246 .
d. weingarten , phys .
* 51 * ( 1983 ) 1830 ; s. nussinov , phys . rev . lett . * 51 * ( 1983 ) 2081 ; e. witten , phys . rev
* 51 * ( 1983 ) 2351 ; c. vafa and e. witten , nucl .
phys . b**234 * * ( 1984 ) 173 .
b. r. zhou , phys .
lett . b**215 * * ( 1988 ) 364 ; b**235 * * ( 1990 ) 412(e ) ; commun .
theor . phys.*15 * ( 1991 ) 319 ; * 18 * ( 1992 ) 347 ; * 19 * ( 1993 ) 377 .
v. a. miransky , _ dynamical symmetry breaking in quantum field theory _ , ( world scientific publishing co.,singapore , 1993 ) and the references therein .
j. c. collins and m. j. perry , phys .
* 34 * ( 1975 ) 1353 .
m. alford , k. rajagopal , and f. wilczek , phys .
lett . b**422 * * ( 1998 ) 427 ; nucl phys .
b**537 * * ( 1999 ) 443 .
r. rapp , t. sch@xmath144fer , e. v. shuryak , and m.velkovsky , phys .
* 81 * ( 1998 ) 53 .
j. berges and k. rajagopal , nucl .
phys . b**538 * * ( 1999 ) 215 .
m. buballa , phys .
* 407 * ( 2005 ) 205 .
d. diakonov , h. forkel , and m. lutz , phys . lett.b
* 373 * ( 1996 ) 147 .
b. vanderheyden and a.d .
jackson , phys .
d**61 * * ( 2000 ) 076004 . s. b. rster and d. h. rischke , phys .
d**69 * * ( 2004 ) 045011 .
j.jaeckel and c. wetterich , phys .
d , * 68 * ( 2003 ) 025020 .
m. asakawa and k. yazaki , nucl .
phys . a**504 * * ( 1989 ) 668 .
y. nambu , phys .
* 117 * ( 1960 ) 648 ; l. p. gorkov , jetp * 7 * ( 1958 ) 993 .
d. ebert and h. reinhardt , nucl .
phys . b**271 * * ( 1986 ) 188 ; b. rosenstein , b. j. warr , and s. h. park , phys .
* 205 * ( 1991 ) 59 ; zhou bang - rong , commun .
* 39 * ( 2003 ) 663 . | by means of a relativistic effective potential , we have analytically researched competition between the quark - antiquark condensates @xmath0 and the diquark condensates @xmath1 in vacuum in ground state of a two - flavor nambu - jona - lasinio ( njl ) model and obtained the @xmath2 phase diagram , where @xmath3 and @xmath4 are the respective four - fermion coupling constants in scalar quark - antiquark channel and scalar color anti - triplet diquark channel .
the results show that , in the chiral limit , there is only the pure @xmath0 phase when @xmath5 , and as @xmath6 decreases to @xmath7 one will first have a coexistence phase of the condensates @xmath0 and @xmath1 and then a pure @xmath1 phase . in non - zero bare quark mass case ,
the critical value of @xmath6 at which the pure @xmath0 phase will transfer to the coexistence phase of the condensates @xmath0 and @xmath8 will be less than 2/3 .
our theoretical results , combined with present phenomenological fact that there is no diquark condensates in the vacuum of qcd , will also impose a real restriction to any given two - flavor njl model which is intended to simulate qcd , i.e. in such model the resulting smallest ratio @xmath6 after the fierz transformations in the hartree approximation must be larger than 2/3 .
a few phenomenological qcd - like njl models are checked and analyzed . |
experiments to search for neutrinoless double beta ( @xmath22 ) decay are considered as a promising way to study properties of neutrino and weak interactions , test the lepton number conservation .
in addition , the process can be mediated by right - handed currents in weak interaction , existence of massless ( or very light ) nambu - goldstone bosons ( majorons ) , and many other effects beyond the standard model @xcite .
experimental efforts are concentrated mainly on the double beta decay with emission of two electrons ( see reviews @xcite ) .
the results of the experiments to search for double beta processes with decrease of nuclear charge : the capture of two electrons from atomic shells ( @xmath23 ) , electron capture with positron emission ( @xmath24 ) , and emission of two positrons ( @xmath25 ) are substantially more modest ( we refer reader to the reviews @xcite and references 1127 in @xcite ) .
even the allowed two neutrino mode of the double beta plus processes is not yet detected unambiguously : there are only indications on two neutrino double electron capture in @xmath26ba @xcite and @xmath27kr @xcite with the half - lives on the order @xmath28 yr . at the same time a strong motivation to search for neutrinoless @xmath24 decay is related with the possibility to refine the mechanism of the @xmath29 decay when observed : whether it appears due to the majorana mass of neutrino or due to the contribution of the right - handed admixtures in the weak interaction @xcite . in addition , experimental data on the two neutrino decay could be useful to improve theoretical calculations of the decay probability .
the isotope @xmath0cd ( energy of decay @xmath30 kev @xcite , natural isotopic abundance @xmath31 @xcite ) is one of the most suitable nuclei to search for the double beta plus processes .
in addition the isotope @xmath0cd is favored to search for resonant @xmath32 transitions to excited levels of @xmath0pd when there is a coincidence between the released energy and the energy of an excited state @xcite .
the decay scheme of @xmath0cd is presented in fig .
[ fig01 ] .
the nuclide @xmath0cd is one of the most studied ( see a detailed review of the previous investigations in @xcite ) . at present
there are three experiments in progress aiming at search for double beta decay of @xmath0cd .
the tgv-2 experiment in the modane underground laboratory in france utilizes 32 planar hpge detectors with 16 thin foils of enriched @xmath0cd ( @xmath33 ) installed between the detectors .
recently new enriched foils with a higher isotopic concentration of @xmath0cd @xmath34 were installed .
the sensitivity of the experiment is at the level of @xmath35 yr @xcite .
the main advantage of the tgv-2 experiment is a high sensitivity to the @xmath36 decay of @xmath0cd with the theoretically shortest half - life of the three allowed channels .
the cobra experiment uses an array of cdznte room temperature semiconductors @xmath37 @xmath38 each at the gran sasso underground laboratory in italy .
the sensitivity of the experiment to the double beta processes in @xmath0cd is at the level of @xmath39 yr @xcite .
the third experiment , also carried out at the gran sasso laboratory , utilizes a radiopure cadmium tungstate crystal scintillator with mass 216 g produced from cadmium enriched in @xmath0cd to 66% ( @xmath0cdwo@xmath1 ) @xcite . at the first stage of the experiment the sensitivity at the level of @xmath40 yr was reached for different channels of double beta decay of @xmath0cd @xcite . to increase the experimental sensitivity to the @xmath7 processes with emission of @xmath41 quanta
, the @xmath0cdwo@xmath1 scintillator was placed inside a low background hpge detector with four ge crystals . here
we report the final results of the experiment with the @xmath0cdwo@xmath1 scintillation detector operated in coincidence ( anticoincidence ) with the four crystals hpge @xmath41 detector .
preliminary results of the experiment were presented in conference proceedings @xcite .
the @xmath0cdwo@xmath1 crystal scintillator was viewed through a lead tungstate ( pbwo@xmath1 ) crystal light - guide ( @xmath42 mm ) by 3 inches low radioactive photomultiplier tube ( pmt ) hamamatsu r6233mod ( see fig . [ fig02 ] ) .
the pbwo@xmath1 crystal was developed from deeply purified @xcite archaeological lead @xcite .
the detector was installed in an ultra - low background gemulti hpge @xmath41 spectrometer of the stella ( subterranean low level assay ) facility @xcite at the gran sasso underground laboratory of the infn ( italy ) at the depth of 3600 m of water equivalent .
four hpge detectors of the gemulti set - up are mounted in one cryostat with a well in the centre .
the volumes of the hpge detectors are approximately 225 @xmath38 each .
an event - by - event data acquisition system is based on two four - channel digital spectrometers ( dgf pixie-4 , xia , llc ) .
one device ( marked ( 1 ) in fig .
[ fig02 ] ) is used to provide spectrometric data for the hpge detectors , while the second pixie-4 ( 2 ) acts as a 14-bit waveform digitizer to acquire signals from the @xmath0cdwo@xmath1 detector at the rate of 18.8 msps over a time window 54.8 @xmath43s .
the second pixie-4 unit records also trigger signals from the home made unit sst-09 , which provides the triggers only if the signal amplitude in the @xmath0cdwo@xmath1 detector exceeds @xmath44 mev to avoid acquisition of a large amount of data caused by the decays of @xmath45cd@xmath46 ( @xmath47 kev ) present in the @xmath0cdwo@xmath1 crystal @xcite .
the signals from the timing outputs of the hpge detectors after summing are fed to the third input of the second pixie-4 digitizer to select coincidence between the @xmath0cdwo@xmath1 and hpge detectors off - line .
the @xmath0cdwo@xmath1 and hpge detectors were calibrated with @xmath48na , @xmath49co , @xmath50cs and @xmath51th @xmath41 sources .
the energy resolution of the @xmath0cdwo@xmath1 detector can be described by the function : fwhm = @xmath52 , where fwhm and @xmath53 are given in kev .
the energy resolution of the hpge detectors during the experiment was fwhm @xmath54 kev for the 1332 kev @xmath41 quanta of @xmath49co .
energy spectrum and distribution of the start positions of the @xmath0cdwo@xmath1 detector pulses relatively to the hpge signals accumulated with the @xmath48na @xmath41 source ( see upper panel in fig .
[ fig03 ] ) demonstrate presence of coincidences between the @xmath0cdwo@xmath1 and hpge detectors under the condition that the energy of events in the hpge detectors is equal to 511 kev ( energy of annihilation @xmath41 quanta ) , while practically there is no coincidence in the data accumulated with @xmath50cs
. as one can see in fig .
[ fig03 ] , the distributions simulated by the egs4 code @xcite are in agreement with the experimental data obtained with @xmath48na , @xmath50cs and @xmath51th gamma sources .
it should be stressed that the energy threshold of the @xmath0cdwo@xmath1 detector in the coincidence mode ( @xmath55 kev ) is much lower than that in the anticoincidence mode since the data acquisition in the coincidence mode is triggered by the signals in the hpge detectors .
the mean - time pulse - shape discrimination method ( see , e.g. , @xcite ) was used to discriminate @xmath56 events from @xmath57 events caused by internal contamination of the crystal by uranium and thorium .
the scatter plot of the mean time versus energy of the background events accumulated over 571 h by the @xmath0cdwo@xmath1 detector is depicted in fig .
[ fig04 ] . the efficiency of the pulse - shape discrimination is worse than that in the previous experiment @xcite due to the lower light collection efficiency with the pbwo@xmath1 light - guide .
nevertheless , the distribution of the mean time for the events with energies in the range @xmath58 mev ( see inset in fig . [ fig04 ] ) justifies pulse - shape discrimination between @xmath57 particles and @xmath41 quanta ( @xmath59 particles ) .
the mean time for @xmath41 quanta was measured with the @xmath51th @xmath41 source in the energy range @xmath60 mev as @xmath61 .
the energy dependence of the mean time distribution sigma ( @xmath62 ) was determined as @xmath63 , where @xmath64 is energy of gamma quanta in kev .
the pulse - shape discrimination was also used to reject overlapped pulses ( mainly caused by pileups of the @xmath45cd@xmath46 @xmath59 decay events ) and pileups of the pbwo@xmath1 scintillation signals ( characterized by rather short decay time at the level of @xmath65s @xcite ) with the @xmath0cdwo@xmath1 signals ( effective average decay time @xmath66s @xcite ) .
the energy spectra accumulated over 13085 h by the @xmath0cdwo@xmath1 detector in anticoincidence with the hpge detectors , in coincidence with event(s ) in at least one of the hpge detectors with energy @xmath67 kev , @xmath68 kev and @xmath69 kev are presented in fig .
[ fig05 ] .
the events in the anticoincidence spectrum were selected by using the following cuts : 1 ) there is no signal(s ) in the hpge detectors , 2 ) the mean time ( @xmath70 ) value of a scintillation pulse is within the plus - minus two sigma interval around the central value : @xmath71 ( the interval is shown in fig .
[ fig04 ] ) .
the pulse - shape discrimination cut selects 95.5% of @xmath56 events .
the spectrum of the @xmath0cdwo@xmath1 detector in coincidence with the energy release in at least one of the hpge detectors of more than 200 kev was built with the help of the same pulse - shape discrimination cut .
the energy spectra of the @xmath0cdwo@xmath1 detector in coincidence with the signal(s ) in the hpge detectors with energy 511 kev ( 1160 kev ) were built by using the following cuts : 1 ) there is event(s ) in at least one of the hpge detectors with energy @xmath72 kev ( @xmath73 kev ) where @xmath74 and @xmath75 are the energy resolutions of the hpge detectors for the annihilation peak and for gamma quanta with energy 1160 kev , respectively ; 2 ) the signals in the @xmath0cdwo@xmath1 and hpge detectors coincide within the plus - minus three sigma time interval ( see inset of fig .
[ fig03 ] , the time cut selects 99.7% of the coinciding events ) ; 3 ) the mean time value of a @xmath0cdwo@xmath1 scintillation pulse is within the @xmath76 interval .
the pulse - shape discrimination cut selects 99.7% of the @xmath56 events .
the data accumulated by the @xmath0cdwo@xmath1 detector in anticoincidence with the hpge detectors confirmed the assumption about surface contamination of the @xmath0cdwo@xmath1 crystal by @xmath77bi @xcite .
the @xmath41 peaks of @xmath77bi observed in @xcite disappeared after the cleaning of the scintillator by potassium free detergent and ultra - pure nitric acid .
the anticoincidence spectrum was fitted in the energy interval @xmath78 kev ( @xmath79/n.d.f.@xmath80 , where n.d.f .
is the number of degrees of freedom ) by the model built from the energy distributions simulated by egs4 code @xcite .
the model includes radioactive contamination of the @xmath0cdwo@xmath1 crystal scintillator @xcite , external gamma quanta from the materials of the set - up ( @xmath81k , @xmath82th , @xmath83u in the cryostat of the hpge detector , pmt , pbwo@xmath1 light - guide , @xmath84al in the aluminium well of the cryostat ) , distribution of alpha particles ( which passed the pulse - shape discrimination cut to select beta and gamma events ) , cosmogenic @xmath85ag@xmath46 in the @xmath0cdwo@xmath1 scintillator and two neutrino @xmath86 decay of @xmath87cd with the half - life @xmath88 yr @xcite present in the @xmath0cdwo@xmath1 crystal with the isotopic abundance @xmath89 @xcite .
the result of the fit and the main components of the background are shown in fig .
[ fig06 ] . the energy spectrum accumulated by the hpge detector
is presented in fig .
[ fig07 ] together with the background data taken over 4102 h. the counting rate of the hpge detector with the @xmath0cdwo@xmath1 detector inside exceeds slightly the background counting rate .
some excess ( at the level of 30% 170% depending on the energy of @xmath41 quanta ) is observed in the peaks of @xmath90bi and @xmath90pb ( daughters of @xmath91ra from the @xmath83u family ) .
we assume that the excess is due to the radioactive contamination of the @xmath0cdwo@xmath1 detector and due to the hpge detector passive shield modification ( some part of the passive shield was removed to install the @xmath0cdwo@xmath1 detector ) .
we have also observed gamma quanta with energy 1809 kev , which we ascribe to cosmogenic @xmath84al in the aluminium well of the cryostat .
besides , there is a clear peak at the energy @xmath92 kev with the counting rate 0.381(4 ) counts / hour ( the peak is shown in the middle part of fig .
[ fig07 ] ) .
we ascribe the peak to gamma quanta of the isomeric transition of @xmath45cd@xmath46 with energy 263.7(3 ) kev @xcite .
the response functions of the @xmath0cdwo@xmath1 and hpge detectors to the @xmath7 processes in @xmath0cd were simulated with the help of the egs4 code .
the simulated energy distributions in the @xmath0cdwo@xmath1 detector without coincidence and in coincidence with 511 kev @xmath41 quanta ( in coincidence with 622 kev gamma quanta for the @xmath93 and @xmath4 decays to the 1134 kev excited level of @xmath0pd ) in the hpge detector are presented in fig .
[ fig08 ] .
there are no peculiarities in the data accumulated with the @xmath0cdwo@xmath1 and hpge detectors that could be ascribed to the @xmath7 processes in @xmath0cd
. therefore only lower half - life limits can be set by using the formula : @xmath94 where @xmath95 is the number of @xmath0cd nuclei in the @xmath0cdwo@xmath1 crystal ( @xmath96 ) , @xmath97 is the detection efficiency , @xmath98 is the time of measurements , and @xmath99 is the number of events of the effect searched for , which can be excluded at a given confidence level ( cl ) .
all the limits are presented in this paper with 90% cl .
we have analyzed different data to estimate limits on the @xmath7 processes in @xmath0cd .
for instance , to derive the value of @xmath99 for the @xmath100 decay of @xmath0cd to the ground state of @xmath0pd , the @xmath0cdwo@xmath1 anticoincidence spectrum was fitted by the model built from the components of the background and the effect searched for .
the best fit , achieved in the energy interval @xmath101 kev , gives the area of the effect @xmath102 counts , thus providing no evidence for the effect . in accordance with the feldman - cousins procedure @xcite
, this corresponds to @xmath103 counts .
taking into account the detection efficiency within the interval given by the monte carlo simulation ( 69.3% ) and the 95.5% efficiency of the pulse - shape discrimination to select @xmath41 and @xmath59 events , we got the half - life limit : @xmath104 yr . the excluded distribution of the @xmath100 decay of @xmath0cd to the ground state of @xmath0pd is shown in fig .
[ fig06 ] . the limit is lower than that obtained in the previous stage of the experiment @xcite due to slightly higher background in the high energy part and worse energy resolution of the @xmath0cdwo@xmath1 detector .
the counting rate of the @xmath0cdwo@xmath1 detector is substantially suppressed in coincidence with the energy 511 kev in the hpge detectors .
the coincidence energy spectrum of the @xmath0cdwo@xmath1 detector is presented in fig .
[ fig09 ] .
there are only 115 events in the energy interval @xmath105 mev , while the simulated background model ( built by using the parameters of the anticoincidence spectrum fit ) contains 108 counts .
we have estimated values of @xmath99 for the @xmath7 processes in @xmath0cd in different energy intervals .
for instance , there are 62 counts in the energy interval @xmath106 kev , while the background model contains 68 counts .
according to @xcite , one should take @xmath107 counts .
taking into account the detection efficiency of the set - up to the @xmath4 decay ( 7.59% ) , the part of the energy spectrum in the energy interval ( 92.3% ) , the selection efficiency of the pulse - shape discrimination , time and energy cuts used to build the coincidence spectrum ( 98.4% in total ) , we have obtained the limit on the @xmath4 decay : @xmath108 yr .
however , this limit depends on the model of background which has some uncertainty related with our limited knowledge of the radioactive impurities composition and localization in the set - up .
in addition , in some energy intervals ( as in the considered above @xmath106 kev ) the measured number of events is less than in the model of background .
the authors of work @xcite suggest in such a situation to report the so called `` sensitivity '' of the experiment defined as the average upper limit that would be obtained by an ensemble of experiments with the expected background and no true signal .
following these suggestions we took a more conservative , background model independent limit on the number of events of the effect searched for as @xmath109 counts , which leads to the following half - life limit on the @xmath4 decay of @xmath0cd to the ground state of @xmath0pd which we accept as a final result : @xmath110 yr .
the excluded distribution of the @xmath4 decay of @xmath0pd to the ground state of @xmath0pd is presented in fig .
[ fig09 ] together with the excluded spectra of the neutrinoless and two neutrino double positron decay of @xmath0cd .
the limits on the processes with positron(s ) emission for transitions to the ground and excited states of @xmath0pd were obtained in a similar way mainly by using the coincidence of the @xmath0cdwo@xmath1 detector with annihilation @xmath41 quanta in the hpge detectors ( the limits are presented in table 1 ) .
a rather high sensitivity was achieved for the @xmath24 and @xmath25 decay channels thanks to the comparatively high probability to detect at least one annihilation gamma quantum by the hpge counters ( e.g. , 13.6% for the @xmath25 decay ) .
further suppression of the background was achieved by selection of events in the @xmath0cdwo@xmath1 detector in coincidence with the intense @xmath41 quanta in the hpge counters expected in the double electron capture in @xmath0cd to the excited levels of @xmath0pd .
for instance , gamma quanta with energy 616 kev expected in the @xmath23 decay of @xmath0cd to the excited level 1128 kev of @xmath0pd can be detected by the hpge counters with a detection efficiency 2.42% ( 2.34% ) in the two neutrino ( neutrinoless ) process . as one can see in fig .
[ fig10 ] , there are only 17 ( 23 ) counts in the energy interval @xmath111 kev ( @xmath112 kev ) . according to @xcite one should take a `` sensitivity '' estimation of @xmath113 counts .
the energy intervals contain 78.2% ( 86.2% ) of the simulated @xmath93 ( @xmath32 ) spectra .
taking into account the efficiency of the pulse - shape discrimination , the time and energy coincidence with the gamma quanta 616 kev ( 97.3% in total ) , the following half - life limit on the @xmath93 ( @xmath32 ) decay of @xmath0cd to the excited level 1128 kev of @xmath0pd can be obtained : @xmath114 yr ( @xmath115 yr ) . the excluded distributions of the @xmath93 and @xmath32 decays of @xmath0cd to the 1128 kev excited level of @xmath0pd are presented in fig .
[ fig10 ] .
most of the limits on the double electron capture in @xmath0cd to the excited levels of @xmath0pd were obtained in a similar way ( including the transitions to the levels of @xmath0pd with energies 2718 kev and 2748 kev where the resonant processes are possible ) .
typical detection efficiencies for the @xmath23 processes in @xmath0cd to the excited levels of @xmath0pd are @xmath116 depending on the level and decay mode , while the number of counts varies from 83 to 0 depending on the energy in the hpge detectors chosen to build the coincidence spectra .
for example , 83 coincidence counts were obtained in the energy interval @xmath117 kev in cc with energy 512 kev in the hpge counters , while 0 events were detected in the energy interval @xmath118 kev in the coincidence with the energy 1160 kev in the hpge detectors ( resonant @xmath119 capture to the 2718 kev excited level of @xmath0pd ) .
we have also used the data accumulated by the hpge detectors to estimate limits on the @xmath7 processes in @xmath0cd .
for instance , in neutrinoless @xmath23 capture we assume that the energy excess is taken away by bremsstrahlung @xmath41 quanta with energy @xmath120 , where @xmath121 is the binding energy of @xmath122-th captured electron on the atomic shell , and @xmath123 is the energy of the populated ( g.s . or excited ) level of @xmath0pd . in case of transition to an excited level , in addition to the initial @xmath41 quantum , other @xmath41 s will be emitted in the nuclear deexcitation process .
for example , to derive a limit on the @xmath119 capture in @xmath0cd to the ground state of @xmath0pd the energy spectrum accumulated with the hpge detectors was fitted in the energy interval @xmath124 kev by a simple function ( first degree polynomial function to describe background plus gaussian peak at the energy 2726.7 kev with the energy resolution fwhm@xmath125 kev to describe the effect searched for ) .
the fit gives an area of the peak @xmath126 counts , with no evidence for the effect .
according to @xcite we took 11.4 events which can be excluded with 90% cl . taking into account the detection efficiency for gamma quanta with energy 2726.7 kev in the experimental conditions ( 1.89% ) we have set the following limit for the @xmath119 capture of @xmath0cd to the ground state of @xmath0pd : @xmath127 yr .
limits on the @xmath128 and @xmath129 capture in @xmath0cd to the ground state of @xmath0pd ( @xmath130 yr and @xmath131 yr , respectively ) were derived in the similar way .
the excluded peaks are shown in fig .
[ fig11 ] .
we would like to stress the certain advantages of isotope @xmath0cd for which the high accuracy @xmath132 value is now available .
thanks to this feature one could distinguish clearly the @xmath133 , @xmath134 and @xmath129 modes of the decay using high resolution hpge detectors .
all the half - life limits obtained in the present work are summarized in table 1 , where results of the most sensitive previous experiments are given for comparison .
.half - life limits on 2@xmath59 processes in @xmath0cd .
the experimental selection ( ac anticoincidence , cc in coincidence at the given energy with hpge , hpge using only the hpge detectors ) is also reported .
the results of the most sensitive previous experiments are given for comparison . [ cols= " < ,
< , < , < , < " , ] [ 2b - results ] an upper limit on the effective nuclear matrix element for the @xmath4 decay ( @xmath135 ) can be obtained with the help of the following formula : @xmath136 where @xmath137 is the phase space factor for the @xmath4 decay , @xmath138 , @xmath139 is the axial vector coupling constant , and @xmath140 is the electron mass . using the recently calculated phase space factors @xmath141 yr@xmath142 @xcite and @xmath143 yr@xmath142 @xcite , we have obtained the following limit on the nuclear matrix element for the @xmath4 decay of @xmath0cd to the ground state of @xmath0pd : @xmath144 .
an experiment to search for @xmath7 decay of @xmath0cd with enriched @xmath0cdwo@xmath1 crystal scintillator in coincidence with four crystals hpge @xmath41 detector has been completed after 13085 h of data taking .
new limits on @xmath23 , @xmath24 and @xmath25 processes in @xmath0cd were set on the level of @xmath145 yr .
the highest sensitivity was achieved mainly by analysis of coincidence between the @xmath0cdwo@xmath1 and the hpge detectors . for the decay channels with emission of positrons ( @xmath24 and @xmath25 decays ) a higher sensitivity was achieved in coincidence with annihilation gamma quanta , while coincidence of the @xmath0cdwo@xmath1 detector with intensive gamma quanta in the hpge counters gives the most stringent limits on the double electron capture in @xmath0cd to excited levels of @xmath0pd .
the half - life limit on the two neutrino @xmath24 decay , @xmath146 yr , reached the region of some theoretical predictions @xcite . by using the half - life limit
the nuclear matrix element for the @xmath4 decay of @xmath0cd to the ground state of @xmath0pd can be bounded as @xmath147 .
the resonant neutrinoless double electron captures to the 2718 kev , 2741 kev and 2748 kev excited states of @xmath0pd are restricted on the level of @xmath148 ) yr .
unfortunately , the experiment has no competitive sensitivity to the most probable @xmath36 channel of the decay due to the high activity of @xmath45cd@xmath46 in the @xmath0cdwo@xmath1 crystal scintillator .
advancement of the experiment is in progress using the @xmath0cdwo@xmath1 detector in coincidence with two large volume cdwo@xmath1 scintillation detectors in close geometry to improve the detection efficiency to gamma quanta emitted in the double beta processes in @xmath0cd .
the authors from the institute for nuclear research ( kyiv , ukraine ) were supported in part by the project co-1 - 2/2015 of the program of collaboration with the joint institute for nuclear research ( dubna ) `` promising basic research on high energy and nuclear physics '' for 2014 - 2015 of the national academy of sciences of ukraine .
we are grateful to s.r .
elliott for his note on presence of @xmath87cd @xmath149 decay in @xmath0cdwo@xmath1 data .
we would like to thank prof .
h. ejiri for useful discussions . j.d .
vergados , h. ejiri , f. @xmath150imkovic , theory of neutrinoless double - beta decay , rep .
phys . * 75 * , 106301 ( 2012 ) .
j. barea , j. kotila , f. iachello , limits on neutrino masses from neutrinoless double-@xmath59 decay , phys .
lett . * 109 * , 042501 ( 2012 ) .
w. rodejohann , neutrino - less double beta decay and particle physics , j. phys .
g * 39 * , 124008 ( 2012 ) .
deppisch , m. hirsch , h. p@xmath151s , neutrinoless double - beta decay and physics beyond the standard model , j. phys .
g * 39 * , 124007 ( 2012 ) .
bilenky , c. giunti , neutrinoless double - beta decay : a probe of physics beyond the standard model , int .
a * 30 * , 1530001 ( 2015 ) .
tretyak , yu.g .
zdesenko , tables of double beta decay data , at .
data nucl .
data tables * 61 * , 43 ( 1995 ) .
tretyak , yu.g .
zdesenko , tables of double @xmath59 decay data an update , at .
data nucl .
data tables * 80 * , 83 ( 2002 ) .
elliott , recent progress in double beta decay , mod .
a * 27 * , 1230009 ( 2012 ) .
a. giuliani , a. poves , neutrinoless double - beta decay , ahep * 2012 * , 857016 ( 2012 ) .
r. saakyan , two - neutrino double - beta decay , annu .
nucl . part .
sci . * 63 * , 503 ( 2013 ) .
o. cremonesi , m. pavan , challenges in double beta decay , ahep * 2014 * , 951432 ( 2014 ) .
g@xmath152mez - cadenas and j. mart@xmath153n - albo , phenomenology of neutrinoless double beta decay , proc . of sci .
* ( gssi14 ) * , 004 ( 2015 ) .
x. sarazin , review of double beta experiments , j. phys .
: conf . ser . *
593 * , 012006 ( 2015 ) .
j. maalampi , j. suhonen , neutrinoless double @xmath154/ec decays , ahep * 2013 * , 505874 ( 2013 ) .
p. belli et al .
, search for double-@xmath59 decay processes in @xmath0cd with the help of a @xmath0cdwo@xmath1 crystal scintillator , phys .
c * 85 * , 044610 ( 2012 ) .
a.p . meshik et al . , weak decay of @xmath26ba and @xmath155ba : geochemical measurements , phys .
c * 64 * , 035205 ( 2001 ) .
m. pujol et al . , xenon in archean barite : weak decay of @xmath26ba , mass - dependent isotopic fractionation and implication for barite formation , geochim .
acta * 73 * , 6834 ( 2009 ) . yu.m .
gavrilyuk et al . , indications of @xmath36 capture in @xmath27kr , phys .
c * 87 * , 035501 ( 2013 ) .
m. hirsch et al . , nuclear structure calculation of @xmath156 , @xmath157 and @xmath158 decay matrix elements , z. phys .
a * 347 * , 151 ( 1994 ) .
m. wang et al . , the ame2012 atomic mass evaluation ( ii ) .
tables , graphs and references , chinese phys .
c * 36 * , 1603 ( 2012 ) . m. berglund , m.e .
wieser , isotopic compositions of the elements 2009 ( iupac technical report ) , pure appl .
chem . * 83 * , 397 ( 2011 ) .
krivoruchenko et al . , resonance enhancement of neutrinoless double electron capture , nucl .
a * 859 * , 140 ( 2011 ) .
d. de frenne , a. negret , nuclear data sheets for a = 106 , nucl .
data sheets * 109 * , 943 ( 2008 ) .
rukhadze et al . , new limits on double beta decay of @xmath0cd , nucl .
a * 852 * , 197 ( 2011 ) .
rukhadze et al . ,
search for double beta decay of @xmath0cd , bull .
phys . * 75 * , 879 ( 2011 ) .
briancon et al . , new search for double electron capture in @xmath0cd decay with the tgv-2 spectrometer , phys . at
* 78 * , 740 ( 2015 ) .
j. ebert et al . ,
current status and future perspectives of the cobra experiment , ahep * 2013 * , 703572 ( 2013 ) .
p. belli et al .
, development of enriched @xmath0cdwo@xmath1 crystal scintillators to search for double @xmath59 decay processes in @xmath0cd , nucl .
instr . meth .
a * 615 * , 301 ( 2010 ) .
tretyak et al . , first results of the experiment to search for @xmath7 decay of @xmath0cd with @xmath0cdwo@xmath1 crystal scintillator in coincidence with four crystals hpge detector , epj web of conferences * 65 * , 01003 ( 2014 ) .
o.g . polischuk et al . , search for @xmath7 processes in @xmath0cd with @xmath0cdwo@xmath1 crystal scintillator , functional materials * 22 * , 135 ( 2015 ) .
danevich et al . ,
search for double beta processes in @xmath0cd with enriched @xmath0cdwo@xmath1 crystal scintillator in coincidence with four crystals hpge detector , aip conf .
* 1686 * , 020006 ( 2015 ) .
tretyak et al .
, new limits on @xmath7 processes in @xmath0cd , to be published in _ proceedings of the taup 2015 conference , september 2015 _ , j. phys . : conf .
( 2016 ) , edited by institute of physics publishing ( iop ) .
boiko et al . , ultrapurification of archaeological lead , inorganic materials * 47 * , 645 ( 2011 ) .
danevich et al . , ancient greek lead findings in ukraine , nucl
a * 603 * , 328 ( 2009 ) .
m. laubenstein et al . , underground measurements of radioactivity , appl . radiat .
isotopes * 61 * , 167 ( 2004 ) .
danevich et al . , development of radiopure cadmium tungstate crystal scintillators from enriched @xmath0cd and @xmath87cd to search for double beta decay , aip conf
. proc . * 1549 * , 201 ( 2013 ) .
nelson et al . ,
the egs4 code system , slac - report-265 ( stanford , 1985 ) . l. bardelli et al . , pulse - shape discrimination with pbwo@xmath1 crystal scintillators , nucl .
instr . meth .
a * 584 * , 129 ( 2008 ) .
l. bardelli et al . , further study of cdwo@xmath1 crystal scintillators as detectors for high sensitivity @xmath7 experiments : scintillation properties and pulse - shape discrimination , nucl . instr . meth .
a * 569 * , 743 ( 2006 ) .
danevich et al . , search for double beta decay of @xmath87cd with enriched @xmath87cdwo@xmath1 crystal scintillators ( aurora experiment ) , to be published in _ proceedings of the taup 2015 conference , september 2015 _ , j. phys .
( 2016 ) , edited by institute of physics publishing ( iop ) .
j. blachot , nucl .
data sheets for a = 113 , nucl .
data sheets * 111 * , 1471 ( 2010 ) .
feldman , r.d .
cousins , unified approach to the classical statistical analysis of small signals , phys .
d * 57 * , 3873 ( 1998 ) .
j. kotila , f. iachello , phase space factors for @xmath159 decay and competing modes of double-@xmath59 decay , phys .
c * 87 * , 024313 ( 2013 ) .
m. mirea , t. pahomi , s. stoica , values of the phase space factors involved in double beta decay , romanian reports in physics * 67 * , 872 ( 2015 ) .
p. belli et al .
, new limits on @xmath25 decay processes in @xmath0cd , astropart . phys . * 10 * , 115 ( 1999 ) .
barabash et al .
, theoretical and experimental investigation of the double beta processes in @xmath0cd , nucl .
a * 604 * , 115 ( 1996 ) .
j. toivanen , j. suhonen , study of several double - beta - decaying nuclei using the renormalized proton - neutron quasiparticle random - phase approximation , phys .
c * 55 * , 2314 ( 1997 ) .
rumyantsev , m.h .
urin , the strength of the analog and gamow - teller giant resonances and hindrance of the @xmath160-decay rate , phys .
b * 443 * , 51 ( 1998 ) .
o. civitarese , j. suhonen , is the single - state dominance realized in double-@xmath59-decay transitions ?
c * 58 * , 1535 ( 1998 ) . j. suhonen , o. civitarese , theoretical results on the double positron decay of @xmath0cd , phys . lett .
b * 497 * , 221 ( 2001 ) . | a radiopure cadmium tungstate crystal scintillator , enriched in @xmath0cd to 66% , with mass of 216 g ( @xmath0cdwo@xmath1 ) , was used to search for double beta decay processes in @xmath0cd in coincidence with four ultra - low background high purity germanium detectors in a single cryostat .
new improved limits on the double beta processes in @xmath0cd have been set on the level of @xmath2 yr after 13085 h of data taking .
in particular , the half - life limit on the two neutrino electron capture with positron emission , @xmath3 yr , has reached the region of theoretical predictions . with this half - life limit
the effective nuclear matrix element for the @xmath4 decay is bounded as @xmath5 .
the resonant neutrinoless double electron captures to the 2718 kev , 2741 kev and 2748 kev excited states of @xmath0pd are restricted at the level of @xmath6 ) yr . * search for @xmath7 decay of @xmath0cd with enriched @xmath0cdwo@xmath1 crystal scintillator in coincidence with four hpge detectors * 1.0 cm * * p. belli@xmath8 , r. bernabei@xmath9 , v.b .
brudanin@xmath10 , f. cappella@xmath11 , v. caracciolo@xmath11 , r. cerulli@xmath11 , d.m .
chernyak@xmath12 , f.a .
danevich@xmath12 , s. dangelo@xmath9 , a. di marco@xmath13 , a. incicchitti@xmath14 , m. laubenstein@xmath11 , v.m .
mokina@xmath12 , d.v .
poda@xmath15 , o.g .
polischuk@xmath16 , v.i .
tretyak@xmath16 , i.a .
tupitsyna@xmath17 0.3 cm @xmath8_infn , sezione di roma `` tor vergata '' , i-00133 rome , italy _
@xmath18_dipartimento di fisica , universit di roma `` tor vergata '' , i-00133 rome , italy _ @xmath10_joint institute for nuclear research , 141980 dubna , russia _ @xmath11_infn , laboratori nazionali del gran sasso , i-67100 assergi ( aq ) , italy _
@xmath12_institute for nuclear research , msp 03680 kyiv , ukraine _
@xmath19_infn , sezione di roma , i-00185 rome , italy
_ @xmath20_dipartimento di fisica , universit di roma `` la sapienza '' , i-00185 rome , italy _ @xmath21_centre de
sciences nuclaires et de sciences de la matire , 91405 orsay , france _ @xmath17_institute of scintillation materials , 61001 kharkiv , ukraine _ 0.5 cm 0.4 cm pacs number(s ) : 29.40.mc , 23.40.-s 0.4 cm _ keywords _ : double beta decay , @xmath0cd , low counting experiment , scintillation detector , hpge detector |
high - precision measurements of the deep inelastic scattering data open a possibility for testing new theoretical ideas in a wide kinematic range of @xmath3 and @xmath5 . in the present paper , we report the results of the application of the analytic approach in qcd @xcite , called the analytic perturbation theory ( apt ) , to the qcd analysis of the experimental data on the @xmath2 structure function .
in contrast to the perturbative running coupling which is spoiled by unphysical singularities in the infrared region at a scale @xmath6 , in the framework of the apt , the analytic coupling has no unphysical singularities .
this fact stimulated applications of the apt and its modifications supporting correct analytic properties for various physical processes ( see refs .
@xcite ) , especially after the generalization of the apt method to the noninteger powers of the running coupling @xcite .
this paper is a continuation of our previous analysis @xcite of the @xmath2 structure function data of the ccfr collaboration @xcite . in the present analysis
, we include in the consideration a set of various experimental @xmath2 data : the cdhs @xcite , scat @xcite , bebc - gargamelle @xcite , bebc - wa59 @xcite , nutev @xcite and chorus @xcite .
note that similar perturbative analysis was performed earlier in ref .
@xcite , but later the new data of the nutev and chorus collaborations were published , and we include these data in the present analysis . in accordance with the result of ref .
@xcite concerning the disagreement of cdhs data with perturbative qcd at small @xmath3 , a cut @xmath7 was applied to their data .
the total number of experimental points is 289 and considerably exceeds the number of points considered before in ref .
it allows us to determine parameters of the form of the @xmath2 structure function more precisely .
the kinematic region of a combined set of data is @xmath8 and @xmath9 gev@xmath10 .
thus , in addition to the previous analysis in ref . @xcite the kinematic region @xmath11 gev@xmath10 is now included .
this region provides a possibility for more precise determination of the parameter @xmath1 and the higher twist ( ht ) contribution . in order to achieve a more direct comparison of the perturbation theory ( pt ) and apt results for fit of a combine set of data
, we do not introduce the normalization factors for the data of an individual collaboration . in sec . 2
, we present the relevant theoretical expressions required for our analysis . in sec . 3 , we give the qcd fit results in the pt and apt approaches and compare the deviation of the apt results from the pt ones with the experimental uncertainties .
summarizing remarks are given in sec .
in this section , necessary expressions for our analysis are written and outline how we use them .
we follow the well known approach based on the jacobi polynomial expansion of structure functions .
this method of solution of the dokshitzer - gribov - lipatov - altarelli - parisi ( dglap ) evolution equation @xcite was proposed in refs .
@xcite and developed for both unpolarized @xcite and polarized cases @xcite .
the main formula of this method allows an approximate reconstruction of the structure function through a finite number of mellin moments @xmath12 here @xmath13 is the ht term , @xmath14 are the jacobi polynomials , @xmath15 contain @xmath16- and @xmath17-dependent euler @xmath18-functions where @xmath19 are the jacobi polynomial parameters , fixed by the minimization of the error in the reconstruction of the structure function .
the perturbative renormalization group @xmath5 evolution of moments is well known ( see , e.g. , refs .
@xcite ) and in the leading order reads as @xmath20^{\nu } } { ~[\alpha _ { s}(q_{0}^{2})]^{\nu}}\ , m_3(n , q^2_0 ) , ~~ { \nu(n)}={{{\gamma_{ns}^{(0),n}}/{2\beta_0 } } } , ~n = 2,~3 , ... \,,\ ] ] where @xmath21 is the qcd running coupling , @xmath22 are the nonsinglet one - loop anomalous dimensions , @xmath23 is the first coefficient of the renormalization group @xmath17-function , and @xmath24 denotes the number of active flavors .
unknown quantity @xmath25 in eq .
( [ evol - pt ] ) could be parameterized as the mellin moments of structure function @xmath2 at some point , @xmath26 : @xmath27 the shape of the function @xmath28 in eq .
( [ jacobi ] ) as well as the parameters @xmath29 , @xmath30 , @xmath31 , @xmath32 in eq .
( [ mellin ] ) , and the scale parameter @xmath1 are found by fit of a combined set of @xmath2-data @xcite .
the detailed description of the fitting procedure could be found in ref .
the target mass corrections are taken into account to the order @xmath33 .
we do not take into account the nuclear effect in the @xmath2 ( see estimations of this effect in refs .
@xcite ) . in our analysis
we use the analytic approach in qcd , the apt @xcite .
this approach gives the possibility of combining the renormalization group resummation with correct analytic properties in @xmath5-variable for physical quantities .
it should be noted that the moments of the structure functions should be analytic functions in the complex @xmath5 plane with a cut along the negative real axis ( see ref .
@xcite for more details ) , the ordinary pt description violates analytic properties . in the framework of the apt ,
the analytic coupling has no unphysical singularities and in the low-@xmath5 domain , instead of rapidly changing @xmath5 evolution , as occurs in the pt case , the apt approach leads to rather slowly changing functions ( see , e.g. , refs .
@xcite ) . in the asymptotic region of large @xmath5
the apt and pt results coincide .
the results of the leading order qcd fit of a combined set of @xmath2-data in the pt and apt approaches are presented in table 1 and figs .
[ graph1 ] [ graph6 ] . in both cases
the pt and apt are considered within the kinematic conditions @xmath53 gev@xmath10 , @xmath54 gev@xmath10 , for number of active flavours @xmath55 and @xmath56 . the values of @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath28 and @xmath47 are considered as free parameters .
[ tab:1 ] [ cols="<,^,^ " , ] as seen from table 1 , the analysis of a combined set of the @xmath2-data allows us to determine with good accuracy values of the parameter @xmath1 .
these values are different for the pt and apt cases , although consistent within errors .
errors in table 1 correspond to @xmath57 .
it should be noted that due to the increased volume of experimental data , in the present analysis the errors for @xmath58 and @xmath59 are some times less than in our previous analysis of ccfr data@xcite in which it was obtained @xmath60 mev and @xmath61 mev .
also , one can see from table 1 that the highest twist contribution is defined much more precisely than in our previous analysis @xcite ( for comparison see fig . 4 in ref .
@xcite and fig .
[ graph6 ] below ) .
let us turn now to fig .
[ graph1 ] , in which the extracted @xmath47-values is shown as a point with the corresponding experimental errors .
one could see this point is in the region of small powers of @xmath62 , @xmath63 , which corresponds effectively to small numbers of the mellin moments and , consequently , to small values of @xmath3 .
figure 2 shows the @xmath2-shape obtained in the apt ( solid line ) and the pt ( dotted line ) cases .
one can see that the result for the apt approach is slightly higher than for the pt one for small @xmath3 , and less for large @xmath3 .
further , we will consider the difference between the pt and apt fit results in comparison with the experimental errors and will try to answer the question : in what kinematic areas the deviation goes beyond the experimental uncertainties of the apt - fit result ?
+ [ graph2 ] [ graph3 ] in fig .
[ graph3 ] , we plot the difference in the @xmath3-shape of the @xmath2 structure function @xmath64 in comparison with the experimental uncertainties for @xmath65 at @xmath66 gev@xmath10 .
gray color shows a corridor of experimental errors for @xmath65 . as can be seen from fig .
[ graph3 ] , the difference @xmath67 is beyond the experimental errors as at small @xmath68 as well at large values of a variable @xmath3 , @xmath69 . at large @xmath3 values the deviation of the results of the pt from the apt exceeds several times the width of a corridor of experimental errors .
+ figure [ graph4 ] shows the difference between the moments in the pt and apt determined by using eqs .
( [ evol - pt ] ) and ( [ evol - apt ] ) , respectively , @xmath70 as a function of the number of the mellin moments ( including non integral value of @xmath71 ) in a wide range of @xmath72 at values @xmath73 ( left panel ) and @xmath74 ( right panel ) . comparing the results shown on the left and
right panels in fig .
[ graph4 ] one can see that in case of small @xmath5-values the difference @xmath75 goes beyond experimental errors at @xmath76 and exceeds several times the width of this corridor of the experimental errors at @xmath77 . for @xmath78
this effect is less pronounced .
let us consider the difference between the pt and apt approaches for the evolution factor , @xmath79 , which is given by @xmath80^{\nu}$ ] for the pt and @xmath81 for the apt , see eqs .
( [ evol - pt ] ) and ( [ evol - apt ] ) , respectively . in this case
we consider the quantity @xmath82 as a function of noninteger @xmath36 at @xmath53 gev@xmath10 .
we present @xmath83 in fig .
[ graph5 ] for values @xmath73 ( left panel ) and @xmath74 ( right panel ) .
the shaded area shows the experimental uncertainties of @xmath84 .
this uncertainty is related with errors of the only parameter @xmath52 ( see table 1 ) . + in general , the consideration of figs .
[ graph2 ] [ graph5 ] shows that the greatest difference between the results of the pt and apt approaches should be expected at low @xmath5 and large numbers of mellin moments @xmath71 , which corresponds to large values of the variable @xmath3 . finally , in fig .
[ graph6 ] the results of the @xmath3 dependence for the higher twists contribution in the pt and apt approaches are presented .
they are in qualitative agreement with each other .
our results are in qualitative agreement with the results of other papers on extraction of a contribution of the highest twists in the structure function @xmath2 @xcite .
accuracy of definition of the function @xmath28 considerably ( a few times ) increased , in comparison with our result @xcite of the analysis of ccfr data . in our previous analysis we noticed that @xmath85 for @xmath86 @xcite . here
we confirm this relation with higher accuracy .
it should be noted that this inequality is in qualitative agreement with the result obtained for the shape of the ht contribution for the non - singlet approach for the @xmath87 structure function @xcite .
one could see from fig .
[ graph6 ] that the opposite inequality takes place for small values @xmath88 : @xmath89 .
this work could be considered as a continuation of our work @xcite , in which for the first time the apt approach was applied to the analysis of the @xmath2 structure function data . in this work
, we used a combined set of data presented by various collaborations .
thus , a new kinematic region of experimental data @xmath90 gev@xmath10 was included in analysis . as a result , the qcd scale parameter @xmath47
, the parameters of the shape of the structure function and the @xmath3-shape of the higher twist contribution were defined with high accuracy .
the shape of this contribution is in qualitative agreement with the results of the previous analysis of the @xmath2 structure function data .
we compared the difference of the results of the pt and apt analysis with the corridor of experimental uncertainties .
we revealed that the most essential difference in the pt and apt results should be expected for small @xmath5 and large numbers of mellin moments , which approximately corresponds to large values of a variable @xmath3 .
it is a pleasure for the authors to thank s. v. mikhailov , o. v. teryaev and v. l. khandramai for interest in this work and helpful discussions .
this research was supported by the jinr - belrffr grant f14d-007 , the heisenberg - landau program 2014 , jinr - bulgaria collaborative grant , and by the rfbr grants ( nos .
12 - 02 - 00613 , 13 - 02 - 01005 and 14 - 01 - 00647 ) .
a. j. buras , _ rev . mod .
phys . _ * 52 * , 199 ( 1980 ) .
v. g. krivokhizhin and a. v. kotikov , _ phys .
_ * 40 * , 1059 ( 2009 ) .
f. j. yndurain , _ the theory of quark and gluon interactions _ , 4th ed .
( springer , 2006 ) . | we apply analytic perturbation theory to the qcd analysis of the @xmath0 structure function considering a combined set of deep inelastic scattering data presented by several collaborations , and extract values of the scale parameter @xmath1 , the parameters of the form of the @xmath2 structure function , and the @xmath3-shape of the higher twist contribution .
we study the difference between the results obtained within the standard perturbative and analytic approaches in comparison with the experimental errors and state that the greatest difference occurs for large @xmath3 .
+ _ keywords : _ quantum chromodynamics ; deep inelastic scattering ; structure functions , running coupling ; higher twists . + pacs nos . : 12.38.-t , 12.38.cy , 12.39.bx , 11.55.hx , 11.55.fv * the qcd analysis of the combined set for the @xmath4 structure function data based on the analytic approach * + |
let us consider a a single degree of freedom confined mechanical system described with phase - space conjugate variables @xmath2 .
suppose it conservative .
for a given motion its hamiltonian function is fixed to a certain value @xmath3 of the energy @xmath4 .
solving for the momentum variable @xmath5 leads to @xmath6 .
we suppose that we are in presence of a confinement of the system where we have periodic motions only . then two types of periodic motions are possible : * _ libration _ : phase trajectory is closed , and then @xmath7 and @xmath5 are periodic functions of time with a same period . * _ rotation _ or _ circulation _ : phase trajectory is not closed , then @xmath5 is periodic function of @xmath7 . for either type ,
introduce the _ action variable _
@xcite @xmath8 where the loop integral is understood as performed over a complete period of libration ( resp .
rotation ) .
this determines by inversion of @xmath9 the function @xmath10 .
now we know that the hamilton - jacobi equation @xmath11 obeyed by the action @xmath12 , with @xmath13 , has a solution of the type @xmath14 ( note that the action @xmath15 should not be confused with the action variable @xmath16 ) .
the time - independent @xmath17 is the _
hamilton characteristic function _ which generates the contact transformation @xmath18 at constant hamiltonian , where @xmath19 is the angle variable , conjugate to @xmath16 .
it follows from the definition of @xmath16 that the period @xmath20 ( resp .
frequency @xmath21 ) of motion at fixed energy is @xmath22 ( resp .
@xmath23 ) , and so the time evolution of the angle variable is linear , with period @xmath20 : @xmath24 note that this equation allows to consider the time as proper to the system at given energy . let us suppose that a series of energy measurements on a mechanical system with one - degree of freedom yields an energy spectrum @xmath25 . in this paper families of corresponding action - angle
coherent states are constructed in view to provide a quantization scheme consistent with that discrete sequence of experimental energies .
the construction is based on a bayesian approach : each family corresponds to a choice of probability distributions , @xmath26 ( prior discrete ) , @xmath27 ( posterior continuous ) such that the mean value of the classical energy with respect to the probability @xmath28 is precisely @xmath1 , up to the addition of a constant independent of @xmath29 .
the formalism @xcite can be viewed as a natural extension of the empirical bohr - sommerfeld rule , @xmath30 , where @xmath31 is the planck constant . we know that this quantization is exact for the motion on the circle ( quantization of the angular momentum ) , and valid in the semiclassical regime . in the deep quantum regime
our approach can be viewed as a viable alternative to the canonical quantization , particularly when the latter is impracticable .
for instance , it yields a satisfactory angle operator .
in section [ csfamily ] we give an overview of a general construction of coherent states labelled by elements of a measure set , their use as a quantizer frame , and their suitability in regard with their phase space content .
we specify this formalism in section [ bayes ] to the action - angle phase space of a confined system by following a bayesian construction mentioned above . in section [ expprob ]
we revisit the question of suitability of these action - angle coherent states .
sections [ partprob ] and [ genprob ] are devoted to the question of choice of probability distributions appropriate for our quantization goals .
section [ timevfr ] is an illustration of our approach by examining a few characteristics , like time evolution , of two types of coherent states for the free rotator .
we end this note in section [ concl ] with some comments about questions raised by our approach and possible generalizations .
let @xmath32 be the phase space of a mechanical system , equipped with its symplectic measure @xmath33 .
actually @xmath34 can be any measure space .
let @xmath35 be some countable set and @xmath36 be an orthonormal system @xmath37 made of elements @xmath38 in the hilbert space @xmath39 , with the positiveness and finiteness constraints @xmath40 let @xmath41 be some separable hilbert space with orthonormal basis @xmath42 in one - to - one correspondence with elements of @xmath36 : @xmath43 .
then the following family of states in the companion @xmath41 , labelled by elements of @xmath32 , @xmath44 refers to the complex conjugate , and obey normalization , @xmath45 , and resolution of the unity @xmath46 . due to these two properties , vectors @xmath47
are named coherent states ( cs ) ( in a wide sense ) .
the resolution of the unity * is * precisely the departure point for the corresponding cs quantization of functions ( or distributions when the latter are properly defined ) on @xmath32 which transforms them into linear operators in @xmath41 : @xmath48 @xmath49 may be considered as cs quantizable if @xmath50 is densely defined in @xmath41 or if the so - called _ lower symbol _ of @xmath50 , @xmath51 is a smooth function on @xmath32 viewed as a topological phase space .
hence , the family @xmath52 , @xmath53 of coherent states offers a certain point of view or frame ( in the right hilbertian geometrical sense ) to analyze , in a non - commutative way , the classical set of points @xmath32 . changing the family corresponds to a change of frame and possibly to an equivalent or non equivalent quantization in regard with specific physical quantities , like energy , action , angle , position ... , and their mutual poisson brackets . besides the fundamental quantizer role played by the cs @xmath47 ,
further criteria are usually requested to qualify the latter as suitable from the classical - quantum relation point of view .
we list here a few of them . for this purpose
, we suppose that the cs @xmath47 depend on a parameter , say @xmath31 , the planck constant , such that the classical limit corresponds to @xmath54 . *
* relative error criterion . * for a semi - bounded quantizable function @xmath49 we define the relative error : @xmath55 where @xmath56 is a constant which is chosen such that @xmath57 for all @xmath58 . hence : _ a cs family is said suitable as a family of quasi - classical states with respect to the function @xmath59 if @xmath60 as @xmath54 .
_ * * time evolution criterion in semi - classical regime .
* let @xmath61 be the hamiltonian of a mechanical system with phase space @xmath32 and let @xmath62 be its cs - quantized counterpart .
we define the time evolution of the probability density of a cs @xmath63 in phase space with respect to the measure @xmath64 as the function @xmath65 its probabilistic nature is directly derived from the resolution of the unity . hence : _ a cs family is said suitable as a family of quasi - classical states with respect to the time evolution ruled by the quantum hamiltonian @xmath62 if the support of the function @xmath66 tends to locate on the classical phase space trajectory @xmath67 with initial condition @xmath68 as @xmath54 . _ * * time evolution stability . * _ a cs family is said temporally stable if under the action of the evolution operator @xmath69 , a coherent state is transformed into another coherent state in the same family , up to possibly a phase factor : _
@xmath70 for more details , see @xcite and references therein .
suppose that measurement on a confined one - dimensional system yields the sequence of values for the energy observable ( up to a constant shift ) : @xmath71 supposing a ( prior ) _ uniform _ distribution on the range of the action variable @xmath16 , we define a corresponding sequence of probability distributions @xmath72 , i.e. @xmath73 , with @xmath74 , obeying the two conditions : @xmath75 where @xmath76 and @xmath77 ( resp .
@xmath78 and @xmath79 ) stand for the rotation ( resp .
libration ) type of motion .
the finiteness condition allows to consider the map @xmath80 as a probabilistic model referring to the discrete data , which might be viewed in the present context as a _
prior distribution _ also .
let @xmath41 be a complex separable hilbert space with orthonormal basis @xmath81 .
@xmath41 is the space of quantum states .
let @xmath82 be a rescaled period of the angle variable and @xmath83 be the action - angle phase space for a rotation ( resp .
libration ) motion with given energies the discrete sequence @xmath84 let @xmath85 be the sequence of probability distributions associated with these energies . we suppose @xmath86 in the rotation case .
one then constructs the family of states in @xmath41 for the rotation or libration motion as the following continuous map from @xmath32 into @xmath41 : @xmath87 where the choice of the real sequence @xmath88 is left to us in order to comply with some if not all criteria previously listed . in both cases
the coherent states @xmath89 * are unit vector : @xmath90 * resolve the unity operator in @xmath41 with respect a measure `` in the bohr sense '' @xmath91 @xcite on the phase space @xmath32 : @xmath92 here , we impose @xmath93 with @xmath94 and letting @xmath95 . hence , if the sequence @xmath96 assumes all its values in @xmath97 , then the angular integral reduces to @xmath98 and the measure @xmath99 becomes the ordinary one on the cylinder .
* allow a `` coherent state quantization '' of classical observables @xmath100 , @xmath101 which is compatible with the energy constraint ( [ twocond ] ) on the posterior distribution @xmath102 .
indeed , the cs quantized version of the classical hamiltonian @xmath103 is diagonal in the basis @xmath104 since it is trivially verified that in both cases the quantum hamiltonian is exactly what we expect : @xmath105 actually , the quantization of any function @xmath106 of the action variable alone yields the diagonal operator : @xmath107 on the other hand , the quantization of any @xmath20 periodic function @xmath108 of the angle variable alone yields the operator : @xmath109_{n n ' } |e_n\rg\lg e_{n'}|\ , , \ ] ] where the matrix elements are formally given by : @xmath110_{n n ' } & = \int_{\r\ , \mathrm{or}\ , \r^+ } d\tj \ , \sqrt{p_n(j)\,p_{n'}(j)}\,\lim_{t \to\infty}\frac{1}{t}\int_{-\frac{t}{2}}^{\frac{t}{2 } } d\gamma \ , e^{-i(\alpha_n - \alpha_{n'})\gamma } \ , f_{\rm ang}(\gamma)\\ & = \left\lbrace \begin{array}{ccc } 0 & \mbox{if } & \alpha_n-\alpha_{n'}\notin \frac{2\pi}{\tau } \z\ , ,
\\ \varpi_{n n'}\ , c_k({f_{\rm ang}};\tau ) & \mbox{if } & \alpha_n-\alpha_{n'}= \frac{2\pi}{\tau } k \in \frac{2\pi}{\tau}\z\ , , \end{array}\right .
\end{aligned}\ ] ] where @xmath111 is the @xmath112th fourier coefficient of @xmath113 , and @xmath114 measures correlation between the two distributions @xmath102 , @xmath115 .
note that the diagonal values are all equal to the average of @xmath108 over one period .
also the infinite matrix can be sparse , even just diagonal , depending on the choice of the @xmath28 and @xmath116 s .
hence , the quantization could at end transform classical observables into a commutative algebra of operators .
an important point is that this cs quantization procedure provides , for a given choice of the sequence @xmath96 , a self - adjoint angle operator @xmath117 corresponding to the angle function @xmath118 defined on the real line as the @xmath20-periodic extension of @xmath119 on the interval @xmath120 .
then @xmath121 for @xmath122 and @xmath123 .
since the relative error function involves lower symbols , let us just consider those for the two particular cases of classical functions , @xmath124 and @xmath125 .
@xmath126 where we observe , in the first case , the appearance of a double averaging using the two bayesian facets , and , in the second case , the presence of a deformation of the fourier series of @xmath108 involving , on one hand , the selection rules @xmath127 , and weights of probabilistic origin on the other hand . the relative error ( [ errorgen ] ) expresses relative deviations of the original classical observables to the above types of averaging involving the two facts of the underlying bayesian duality .
the action - angle phase space representation of a particular coherent state @xmath128 , as a function of @xmath129 , is the `` normalized '' overlap @xmath130 hence , the map @xmath131 represents a localization probability distribution , namely a generalized version of the husimi distribution , on the phase space provided with the pseudo - measure @xmath99 .
indeed , the resolution of the identity gives immediately @xmath132 if we choose instead a specific realization of the hilbert space @xmath41 , like that one generated by eigenfunctions of the quantum hamiltonian @xmath62 in `` @xmath7 '' or `` configuration '' representation , @xmath133 , the corresponding representation of the state @xmath134 reads as @xmath135 with corresponding probability density of localization on the range of the @xmath7-variable given by @xmath136 since the cs quantized version @xmath137 of the classical hamiltonian @xmath138 is diagonal in the basis @xmath139 , the time evolution of the cs in both representations is given respectively by @xmath140 snapshots of the time evolution of the corresponding probability densities @xmath141 and @xmath142 are necessary in order to discriminate suitable functions @xmath28 and sequences @xmath96 from the classical limit viewpoint .
note that temporal evolution stability is granted with the choice @xmath143 .
the quantities that are left undetermined in our construction of the cs are the discretely indexed probability distribution @xmath145 and the sequence @xmath146 .
two simplest situations are helpful in giving some hints : the free rotator ( mass @xmath147 on circle of radius @xmath148 ) and the harmonic oscillator ( frequency @xmath149 ) , for which the energies are respectively @xmath150const . and @xmath151const . in the first case ,
a familiar solution @xcite is family of gaussians centered at each integer , with dimensionless width parameter @xmath152 or @xmath153 : @xmath154 with @xmath155 .
this gives the eigenvalues @xmath156 of @xmath157 and @xmath158 the constant shift being the average value of the classical energy with respect to the distribution @xmath159 . by introducing the compton length @xmath160 of the particle eq .
( [ freerotene ] ) reads @xmath161 . as an illustration of the suitability of this gaussian choice
, we show in figure [ specg ] the respective behaviors of the spectrum and lower symbol of the angle operator @xmath117 for different values of the parameter @xmath162 .
, when @xmath163 and @xmath164 .
for the spectrum , @xmath165 . for the lower symbol : @xmath166 ( the lower symbol has little dependence on @xmath16 ) .
one notices the tendency to the sawtoothed behavior of the classical angle function as @xmath167 , i.e. as @xmath168.,title="fig:",width=288 ] , when @xmath163 and @xmath164 . for the spectrum , @xmath165 . for the lower symbol : @xmath166 ( the lower symbol has little dependence on @xmath16 ) .
one notices the tendency to the sawtoothed behavior of the classical angle function as @xmath167 , i.e. as @xmath168.,title="fig:",width=288 ] in the second case , another familiar solution is the discretely indexed gamma distribution : @xmath169 this gives @xmath170 and @xmath171 , the constant shift being the average value of the classical energy with respect to the distribution @xmath159 .
given the classical relation @xmath172 between action variable and energy , and the observational or computed sequence @xmath173 , the central question is to find the sequence of probability distributions @xmath27 ( at least with a satisfying approximation ) , which obey the two fundamental conditions ( [ twocond ] ) . for the rotation case , the departure point could be a normal - like law , possibly modified along a perturbation scheme with expansion parameter the strength of the potential energy @xmath174 .
similarly , in the libration case , the departure could be a gamma - like law , possibly modified along the same lines .
a nice and manageable model is the simple pendulum , whose the quantum version is well known from the solution of the mathieu equation @xcite .
starting from the computed mathieu eigenvalues @xmath1 , an empirical approach consists in starting from the sequence of computed action variables @xmath175 from eq .
( [ action1 ] ) and to impose , in the rotation case , the sequence of normal laws centered at these @xmath176 .
@xmath177 by `` adjusting '' @xmath178 in order to suitably approximate the @xmath1 s with the computed quantities @xmath179 note that with this choice , the eigenvalues @xmath180 of @xmath157 are precisely the @xmath176 s . handling the libration case is more delicate .
another empirical approach consists in introducing the gamma - like distribution : @xmath181 with an auxiliary sequence @xmath182 such that the corresponding moment problem has a solution with a positive measure @xmath183 , @xmath184 that the quantization conditions involving the computed mathieu eigenvalues , @xmath185 are fulfilled ( at least approximately ) .
we note that , by construction , we have @xmath186 , that the action variable @xmath16 is the mean value of @xmath187 with respect to the poisson - like distribution @xmath188 : @xmath189 , and that the quantum action @xmath157 has eigenvalues @xmath190 .
with the gaussian choice ( [ gaussrot ] ) and for a general choice of a sequence of frequencies @xmath116 , the coherent states for the free rotator read @xmath191 the normalization function @xmath192 is given in two forms : @xmath193 and satisfies @xmath194 . in the following we consider two choices of @xmath195 and investigate
what kind of properties of cs is satisfied for each case .
this choice renders exact the quantization of the classical canonical commutation rule @xmath197 .
indeed , we have @xmath198 . concerning the phase space distribution ( [ evphasdist ] ) for time evolution , we obtain the following upper bound : @xmath199 where @xmath200 , @xmath201 , @xmath202 , and @xmath203 . from this
is derived the estimate on the semi - classical behavior at large @xmath204 : @xmath205 for @xmath206 and vanishes if @xmath207 or @xmath208 .
this choice is appropriate to temporal evolution stability @xmath210 this time , the phase space distribution ( [ evphasdist ] ) is bounded as follows .
@xmath211 @xmath212 where @xmath213 .
hence the estimate on semi - classical behavior of @xmath214 at large @xmath204 : @xmath215 for @xmath216 and vanishes if @xmath207 or @xmath217 .
periodic and more generally integrable systems provide a variety of such families of action - angle coherent states .
a future issue is about the best choice of probability distributions @xmath218 .
a more fundamental concerns the physical ( in terms of physical measurement ) equivalence between different families of coherent states from a quantization point of view . finally , the extension of the probabilistic approach presented in this work to unconfined systems and subsequent continuous spectra is possible @xcite . | for a single degree of freedom confined mechanical system with given energy , we know that the motion is always periodic and action - angle variables are convenient choice as conjugate phase - space variables .
we construct action - angle coherent states in view to provide a quantization scheme that yields precisely a given observed energy spectrum @xmath0 for such a system .
this construction is based on a bayesian approach : each family corresponds to a choice of probability distributions such that the classical energy averaged with respect to this probability distribution is precisely @xmath1 up to a constant shift .
the formalism is viewed as a natural extension of the bohr - sommerfeld rule and an alternative to the canonical quantization . in particular
, it also yields a satisfactory angle operator as a bounded self - adjoint operator . |
the dimer - atom scattering process was first solved by skorniakov and ter - martirosian in 1956 @xcite , in the context of three - nucleon scattering , i.e. a two - body bound state between a neutron and a proton ( called a deuteron ) is scattering with a neutron .
they considered equal - mass particles , and found that while the born approximation gives @xmath2 for the scattering length between an ab dimer and an a atom , where @xmath3 is the two - body scattering length between a and b particles , the exact result turns out to be @xmath4 @xcite .
this problem has recently been analyzed with the diagrammatic @xmath0-matrix approach @xcite , in the context of cold quantum gases ( see also @xcite in a different context ) , and the result is in perfect agreement with the earlier results . in the same context , the three - body scattering problem has also been generalized to unequal - mass particles , and analyzed both in real space through solving the three - body schrdinger equation @xcite and in momentum space through the diagrammatic @xmath0-matrix approach @xcite , with perfect agreement in between for all mass ratios .
when the mass ratio of the heavy particles to the light one is below @xmath5 , it is also well - established that the three - body scattering problem is universal , and that the dimer - atom scattering length is proportional to the two - body scattering length @xmath3 ( just like the equal mass case ) with the proportionality factor depending only on the masses of the constituent particles @xcite . however , since three - body efimov bound states emerges for larger mass ratios @xcite , this problem is not universal , and an additional parameter coming from the short - range ( or large - momentum ) three - body physics is needed for an accurate description . here
, we use the diagrammatic @xmath0-matrix approach to analyze the three - body scattering problem between two identical fermions and a third particle .
our results are relevant to the quantum phases of two - species fermi - fermi @xcite . and
bose - fermi @xcite mixtures of atomic gases in the molecular limit .
the remaining paper is organized as follows . after deriving the skorniakov and ter - martirosian integral equation generalized for unequal - mass particles in sec .
[ sec : t - matrix ] , we numerically solve the resultant equation for all mass ratios in sec .
[ sec : a_da ] , and list the exact dimer - atom scattering lengths for all available two - species fermi - fermi and bose - fermi mixtures in sec .
[ sec : mixtures ] .
a brief summary of our conclusions is given in sec .
[ sec : conclusions ] .
in this paper , we are interested in the three - body ( dimer - atom ) s - wave scattering between two identical fermions ( refer to it as a - type particles ) and a third particle ( refer to it as a b - type particle which could be a different species of fermion or a boson ) . in particular , we consider a zero - ranged attractive interaction between a and b particles , and assume there is a weakly bound resonance between them with the binding energy @xmath6 , so that we want to study the scattering between this bound state ( refer to it as an ab dimer ) and the remaining a atom .
detailed description of the diagrammatic @xmath0-matrix approach for the dimer - atom scattering process can be found in the literature for equal - mass particles @xcite , and here we give details of our calculation @xcite for the case of unequal - mass particles .
[ htb ] we begin our analysis by describing the zero temperature diagrammatic representation for the dimer - atom scattering @xmath0-matrix @xmath7 as illustrated in fig . [ fig : t - matrix ] , where @xmath8 and @xmath9 are the incoming and outgoing momenta , and @xmath10 and @xmath11 are the frequencies for the incoming a particle and ab dimer , respectively .
here , we set the center - of - mass momentum of the dimer - atom system to zero , and @xmath12 is the total frequency where @xmath13 is the binding energy of the two - body bound state between a and b particles , and @xmath14 and @xmath15 are masses of the ab dimer and twice the reduced mass of the @xmath16 and @xmath17 particles , respectively . in fig .
[ fig : t - matrix ] , single lines represent retarded free propagators for the a and b particles @xmath18 where @xmath19 is the energy , and @xmath20 is the chemical potential of the corresponding particle . similarly , double lines represent the retarded propagator for the ab dimer , and in three dimensions it can be approximated by a simple one - pole structure @xmath21^{1/2}},\ ] ] which reflects the presence of a two - body bound state between a and b particles .
this dimer propagator is obtained from the resummation of the ab polarization bubbles leading to @xmath22 , $ ] where @xmath23 is the strength of the bare interaction between a and b particles , and the ab polarization bubble is @xmath24 at zero temperature , @xmath25 in three dimensions .
integration over the internal momentum @xmath26 and frequency @xmath27 leads to @xmath28 which in combination with the definition of the two - body scattering length @xmath29 , and the two - body scattering @xmath0-matrix @xmath30 , $ ] between a and b particles , give the final result described in eq . ( [ eqn : dimer - propagator ] ) . in the following , we set @xmath31 since we are interested in the dimer - atom scattering in vacuum
however , note that our diagrammatic calculation for the scattering parameters of the three - body problem is exact , and they are sufficient to describe ultracold quantum gases , since experiments are always performed at low densities .
the calculation of scattering parameters in the presence of many other particles ( arbitrary density ) is much more difficult , and it is not discussed here . in addition , on the right hand side of fig .
[ fig : t - matrix ] , note that the first diagram represents a fermion exchange process , i.e. first the particle b breaks up with the particle a and then it forms a new ab dimer with the remaining a particle .
this is the simplest process contributing to the dimer - atom scattering , e.g. born approximation , and all other ( infinitely many ) possible processes are included in the second diagram . in analytical form ,
the dimer - atom @xmath0-matrix @xmath7 satisfies the following integral equation @xmath32 the minus signs on the right hand side is due to fermi - dirac statistics , i.e. exchanging a fermion brings a minus sign unlike that of a boson .
the integration over frequency @xmath27 can be easily performed by closing the integration contour in the upper half - plane , where both @xmath33 and @xmath34 are analytic functions of @xmath27 , and only a simple pole contribution comes from @xmath35 .
note that this property of @xmath33 is due to the form of eq .
( [ eqn : t - matrix ] ) itself .
this integration sets @xmath36 , and we set @xmath37 in order to have the same frequency dependence for the @xmath0-matrix on both sides of eq .
( [ eqn : t - matrix ] ) .
this leads to @xmath38},\end{aligned}\ ] ] where we redefine the @xmath0-matrix @xmath39 , and @xmath40 is the total energy and @xmath41 is twice the reduced mass of an ab dimer and an a particle .
since we are interested in the zero - ranged low - energy s - wave scattering , we first average out directions of the incoming momentum @xmath8 , and then of the outgoing momentum @xmath9 , leading to @xmath42 for three - dimensional systems , where @xmath43 is the angular - averaged @xmath0-matrix . to obtain the three - body ( dimer - atom ) s - wave scattering length @xmath44 ,
the total energy @xmath45 should be set to the binding energy @xmath46 of the two - body bound state , in the limit of vanishing incoming and outgoing momentum and frequency , i.e. @xmath47 .
this motivates us to introduce the dimer - atom scattering function @xmath48 for which the integral equation becomes @xmath49 in the limit when @xmath50 . here
, the dimer - atom scattering length is @xmath51 , where @xmath52 t_k^{da}(p ) , \label{eqn : a - scattering}\ ] ] gives the full momentum dependence of the dimer - atom scattering function .
the integral equation shown in eq .
( [ eqn : ieqn ] ) as well as the scattering function expression shown in eq .
( [ eqn : a - scattering ] ) reduce to the known results for the equal - mass particles case @xcite when @xmath53 .
since only the fermion exchange process is taken into account in the born approximation , and that neglecting the second term on the right hand side of eq .
( [ eqn : ieqn ] ) leads to @xmath54 , which is consistent with the many - body results @xcite .
however , we need to include both terms and solve the integral equation in order to find the exact dimer - atom scattering length . next , we solve numerically the integral equation given in eq .
( [ eqn : ieqn ] ) as a function of the mass ratio @xmath55 of the constituent particles of the dimer @xcite . for this purpose
, it is convenient to change the upper integration limit to a finite value by a change of variables , e.g. @xmath56 and @xmath57 where @xmath58 .
the resultant integral is calculated by using the gaussian - legendre quadrature method , and using this discretization , we reduce the integral equation to a matrix - eigenvalue problem . [ htb ] the exact solutions and the born approximation values of @xmath51 are shown in fig . [
fig : a_da ] .
when @xmath59 , we find @xmath4 , which is in agreement with the results previously found for equal - mass particles @xcite . the scattering length @xmath44 increases ( decreases ) from this value with increasing ( decreasing ) mass ratio , and @xmath60 in the limit of @xmath61 as expected .
these results exactly match the few - body results of petrov @xcite .
it is quite remarkable that the diagrammatic @xmath0-matrix approach exactly recovers the few - body results for all mass ratios , since the diagrammatic approach is performed in momentum space , while the few body approach is performed in real space @xcite .
note that the born approximation values for @xmath44 are not in agreement with the exact values for any mass ratio , and that the disagreement increases rapidly with increasing mass ratio , but the general qualitative trend is captured by the born approximation as can be seen in fig .
[ fig : a_da ] . [ htb ] in fig .
[ fig : a_da_k ] , we show the scattering function @xmath62 as a function of the outgoing momentum @xmath63 for some mass ratios @xmath55 . when @xmath59 , this is a monotonically decreasing ( positive ) function of @xmath63 with a long tail .
however , beyond some critical mass ratio , this behavior changes dramatically .
for instance , when @xmath64 , after an initial rapid drop , @xmath62 function changes sign and becomes negative at @xmath65 .
beyond this critical momentum , it reaches a minimum value and slowly decays to zero from the negative side . as the mass ratio gets lower ( higher ) ,
@xmath62 changes sign at higher ( lower ) momenta .
for instance , when @xmath66 , corresponding to a three - body system consisting of one @xmath67li atom and two @xmath68k atoms , it changes sign at least once at @xmath69 ( but it changes sign at least once at @xmath70 when @xmath71 ) .
we find clear sign changes when @xmath72 , but we donot know whether @xmath62 changes sign for even lower mass ratios since precision issues obscure the results . although our calculation is not reliable for mass ratios above @xmath5 @xcite , for illustration purposes , in fig .
[ fig : a_da_k](b ) , we show @xmath73 $ ] as a function of @xmath63 when @xmath74 . in this case , the scattering function first changes sign and becomes negative at @xmath75 , but then it changes sign again and becomes positive at @xmath76 . since the large-@xmath63 behavior of this function is again obscured due to precision issues , as can be seen in the figure , we could not resolve whether it has more sign changes at larger @xmath63 values . however , we mention that as the mass ratio increases further to @xmath77 , @xmath62 changes sign at least 4 times , and the oscillating pattern becomes apparent .
this effect could be related to the `` fall of a particle to the center '' phenomenon @xcite , and to the emergence of three - body efimov bound states @xcite , which are known to occur when the mass ratio is large .
when the heavy a particles are separated from each other by a distance @xmath78 ( or equivalently @xmath79 ) , it is known that an exchange of a light b particle mediates an effective @xmath80 attraction between a particles . here
, the coefficient @xmath81 increases with increasing mass ratio @xcite .
since a particles are fermions , this attraction competes with the pauli repulsion which manifests itself as a centrifugal @xmath82 potential , where @xmath83 is the angular momentum .
therefore , when the masses are comparable ( @xmath84 ) , the pauli repulsion is about one order of magnitude stronger than the mediated attraction . when the mass ratio @xmath55 reaches a critical value , i.e. when @xmath85 , the effective interaction ( mediated attraction plus centrifugal repulsion ) between a particles vanishes .
for the lowest p - wave angular momentum channel @xmath86 , this occurs when @xmath87 . beyond this critical mass ratio ,
there is a second critical mass ratio beyond which the effective attraction between a particles is strong enough that the a particles stay in an infinitely small region around each other , i.e. the remaining a particle falls to the center of the attraction @xcite .
this second critical mass ratio can be approximated by @xmath88 @xcite .
for the lowest p - wave angular momentum channel @xmath86 , this occurs when @xmath89 .
note that , since the b particle is already bound to one of the a particles , `` fall of a particle to the center '' is related to emergence of three - body efimov bound states .
more accurate calculations show that the latter occurs when @xmath90 @xcite .
having presented the diagrammatic @xmath0-matrix approach for the three - body ( s - wave ) dimer - atom scattering , for all mass ratios @xmath55 , next we discuss the relevance of our results to the quantum phases of two - species fermi - fermi and bose - fermi mixtures of atomic gases at ultracold temperatures .
it has been shown that @xcite ( see @xcite for experimental confirmation ) , in the strong - attraction or molecular limit , two - species fermi - fermi mixtures with population imbalance can be well - described by effective bose - fermi models , where fermion - fermion pairs behave as molecular bosons ( dimers ) and interact weakly with each other and with the remaining unpaired ( excess ) fermions .
these simpler models only require accurate scattering lengths between two molecular bosons ( dimer - dimer , i.e. @xmath91 ) , and between a molecular boson and an unpaired fermion ( dimer - atom , i.e. @xmath44 ) .
note that the exact three- and four - body results in vacuum are sufficient to describe ultracold atomic mixtures in the molecular limit , since experiments are always performed at low densities .
several fermionic atoms ( @xmath92li , @xmath68k , @xmath93sr @xcite , and @xmath94yb @xcite ) are being currently investigated , and experimental methods for studying two - species fermi - fermi mixtures are being developed in several groups , e.g.@xmath92li-@xmath68k mixture @xcite . thus , anticipating future experiments involving various other mixtures , in table [ table : ff ] , we list the exact dimer - atom scattering lengths for all possible mixtures .
[ htb ] .[table : ff ] the exact scattering length @xmath51 between a bosonic ab dimer ( consisting of one fermionic a atom and one fermionic b atom ) and a fermionic a atom is shown for a list of two - species fermi - fermi mixtures . here
, @xmath44 is in units of the two - body scattering length between a and b particles @xmath3 , and the mass ratios @xmath55 are shown inside the parenthesis @xcite . [ cols="^,^,^,^,^,^",options="header " , ] as we pointed out above , the three - body scattering problem is universal for mass ratios below @xmath5 , and therefore , the dimer - atom scattering is proportional to the two - body scattering length @xmath3 with the proportionality factor depending only on the masses of the constituent particles .
however , since three - body efimov bound states emerges for larger mass ratios @xcite , this problem is not universal ( an additional parameter coming from the short - range three - body physics is needed for an accurate description ) , and our analysis does not include this non - universal effect @xcite . in particular , the @xmath67li-@xmath93sr , @xmath67li-@xmath94yb and @xmath67li-@xmath95yb fermi - fermi mixtures and @xmath96li-@xmath94yb , and @xmath96li-@xmath95yb bose - fermi mixtures have mass ratios that are above the critical ratio for the emergence of three - body efimov bound states .
in this paper , we used the diagrammatic @xmath0-matrix approach to analyze the three - body scattering problem between two identical fermions and a third particle .
the third particle could be a different species of fermion or a boson .
we calculated the exact s - wave dimer - atom scattering length for all mass ratios , and our results exactly match the few - body results of petrov who obtained them by solving the three - body schrdinger equation @xcite .
it is quite remarkable that the diagrammatic @xmath0-matrix approach exactly recovers the few - body results for all mass ratios , since the diagrammatic approach is performed in momentum space , while the few - body approach is performed in real space @xcite .
we also discussed the relevance of our results to the quantum phases of two - species fermi - fermi and bose - fermi mixtures of atomic gases at ultracold temperatures .
in particular , in the strong - attraction or molecular limit , these mixtures can be well - described by simpler effective models , where paired atoms behave as dimers and interact weakly with each other and with the remaining unpaired atoms .
these effective descriptions require only the scattering lengths between two dimers , and between a dimer and an unpaired atom .
anticipating future experiments , we listed the exact dimer - atom scattering lengths for all available two - species fermi - fermi and bose - fermi mixtures .
in addition , we showed that , unlike that of the equal - mass particles case where the three - body scattering @xmath0-matrix decays monotonically as a function of the outgoing momentum , after an initial rapid drop , this function changes sign and becomes negative at large momenta and then decays slowly to zero when the mass ratio of the fermions to the third particle is higher than a critical value ( around @xmath1 ) .
as the mass ratio gets higher , modulations of the @xmath0-matrix become more apparent with multiple sign changes .
we argued that this effect could be related to the `` fall of a particle to the center '' phenomenon @xcite and to the emergence of three - body efimov bound states @xcite .
the author thanks p. pieri for useful discussions , and the scientific and technological research council of turkey ( tb@xmath97tak ) for financial support
. 99 g. v. skorniakov and k. a. ter - martirosian , zh .
fiz . * 31 * , 775 ( 1956 ) ; and sov .
jetp * 4 * , 648 ( 1957 ) .
i. v. brodsky , m. yu .
kagan , a. v. klaptsov , r. combescot , and x. leyronas , phys .
a * 73 * , 032724 ( 2006 ) .
j. levinsen and v. gurarie , phys .
a * 73 * , 053607 ( 2006 ) .
p. f. bedaque and u. van kolck , physics letters b * 428 * , 221 ( 1998 ) .
d. s. petrov , phys .
a * 67 * , 010703(r ) ( 2003 ) .
m. iskin and c. a. r. s de melo , phys .
a * 77 * , 013625 ( 2008 ) . v. n. efimov , yad .
* 12 * , 1080 ( 1970 ) ; sov .
* 12 * , 589 ( 1971 ) ; nucl .
a * 210 * , 157 ( 1973 ) .
o. i. kartavtsev and a. v. malykh , j. phys .
b * 40 * , 1429 ( 2007 ) .
d. s. petrov , c. salomon , and g. v. shlyapnikov , j. phys .
b : at . mol .
* 38 * , s645 ( 2005 ) .
m. taglieber , a .- c .
voigt , t. aoki , t. w. hnsch , and k. dieckmann , phys .
lett . * 100 * , 010401 ( 2008 ) .
e. wille , f. m. spiegelhalder , g. kerner , d. naik , a. trenkwalder , g. hendl , f. schreck , r. grimm , t. g. tiecke , j. t. m. walraven , s. j. j. m. f. kokkelmans , e. tiesinga , and p. s. julienne , phys .
lett . * 100 * , 053201 ( 2008 ) .
voigt , m. taglieber , l. costa , t. aoki , w. wieser , t. w. hnsch , and k. dieckmann , phys .
lett . * 102 * , 020405 ( 2009 ) .
f. m. spiegelhalder , a. trenkwalder , d. naik , g. hendl , f. schreck , and r. grimm , phys .
lett . * 103 * , 223203 ( 2009 ) .
t. g. tiecke _
_ , arxiv:0908.2071 ( 2009 )
. f. m. spiegelhalder _
et al . _ ,
arxiv:1001.5253 ( 2010 ) .
s. inouye , j. goldwin , m. l. olsen , c. ticknor , j. l. bohn , and d. s. jin , phys .
lett . * 93 * , 183201 ( 2004 ) .
s. ospelkaus , c. ospelkaus , l. humbert , k. sengstock , and k. bongs , phys .
* 97 * , 120403 ( 2006 ) . m. zaccanti , c. derrico , f. ferlaino , g. roati , m. inguscio , and g. modugno , phys .
a * 74 * , 041605(r ) ( 2006 ) .
j. j. zirbel , k .- k .
ni , s. ospelkaus , j. p. dincao , c. e. wieman , j. ye , and d. s. jin , phys .
rev . lett . * 100 * , 143201 ( 2008 ) .
m. iskin and c. a. r. s de melo , phys .
. lett . * 97 * , 100404 ( 2006 ) ; phys .
a * 76 * , 013601 ( 2007 ) .
our results are accurate only for mass ratios that are smaller than the critical value @xmath5 , beyond which an additional parameter coming from the short - range ( or large - momentum ) three - body physics is needed for an accurate description @xcite , which is beyond the scope of this paper .
two methods are equivalent because all of the possible scattering processes are taken into account exactly in our diagrammatic approach at zero temperature .
l. d. landau and e. m. lifshitz , _ quantum mechanics _ , pp .
114 ( butterworth - heinemann , oxford , 1999 ) .
( 35.2 ) and the discussion above eq . ( 35.10 ) for the two - body scattering problem in @xcite .
p. pieri and g. c. strinati , phys .
* 96 * , 150404 ( 2006 ) .
yong - il shin , andr schirotzek , christian h. schunck , and wolfgang ketterle , phys .
* 101 * , 070404 ( 2008 ) .
r. grimm , private communication .
t. fukuhara , y. takasu , m. kumakura , and y. takahashi , phys .
* 98 * , 030401 ( 2007 ) .
k. helfrich , h .- w .
hammer , and d.s .
petrov , arxiv:1001.4371 ( 2010 ) . | we use the diagrammatic @xmath0-matrix approach to analyze the three - body scattering problem between two identical fermions and a third particle ( which could be a different species of fermion or a boson ) .
we calculate the s - wave dimer - atom scattering length for all mass ratios , and our results exactly match the results of petrov .
in particular , we list the exact dimer - atom scattering lengths for all available two - species fermi - fermi and bose - fermi mixtures .
in addition , unlike that of the equal - mass particles case where the three - body scattering @xmath0-matrix decays monotonically as a function of the outgoing momentum , we show that , after an initial rapid drop , this function changes sign and becomes negative at large momenta and then decays slowly to zero when the mass ratio of the fermions to the third particle is higher than a critical value ( around @xmath1 ) .
as the mass ratio gets higher , modulations of the @xmath0-matrix become more apparent with multiple sign changes , related to the `` fall of a particle to the center '' phenomenon and to the emergence of three - body efimov bound states . |
giant flares are the most spectacular manifestations of soft gamma repeaters ( sgrs ) . their time histories are characterized by a very rapid ( @xmath1 1 ms ) rise to an intense peak lasting several hundred milliseconds , followed by a weaker , oscillatory phase which exhibits the period of the neutron star .
the spectrum of the peak is very hard and extends to mev energies .
the most energetic giant flare to date is that of 2004 december 27 from sgr1806 - 20 , with an isotropic energy of well over 10@xmath2 erg .
( hurley et al . 2005 ,
palmer et al .
2005 , mereghetti et al .
2005 , terasawa et al .
2005 , frederiks et al .
the observation of this event raised the question once more of the existence of extragalactic giant magnetar flares , and their relation to the short cosmic gamma - ray bursts ( grbs ) .
viewed from a large distance , only the initial peak of a giant flare would be detectable , and it would resemble a several hundred millisecond long , hard spectrum grb .
the energetics of giant flares make it a virtual certainty that such events can be detected in distant galaxies , but recognizing them and demonstrating their origin beyond a reasonable doubt remain difficult tasks .
predictions of the rates of extragalactic giant flares , and the percentage of short cosmic bursts which might actually be giant flares , vary widely ( from a few to @xmath3 15% - lazzati et al .
2005 , nakar et al .
2006 , popov and stern 2006 , tanvir et al .
2005 ) , in part because of their unknown number - intensity relation ( no sgr has yet been observed to emit more than one giant flare , and their distances are uncertain in most cases ) .
however , they generally agree that the rate is small enough that the majority of short bursts are indeed not due to sgr giant flares . on the other hand , the rate is not expected to be zero , so it is important to examine all possible cases exhaustively . in this paper
, we revisit grb 051103 , a short burst whose location , time history , and energy spectrum are consistent with an origin as a giant flare in m81 .
we make use of the full interplanetary network ( ipn ) data set to obtain a refined localization ( an error ellipse ) .
we utilize the capability of the rhessi spacecraft to obtain time - resolved energy spectra with good energy resolution , at a time resolution which is limited only by counting statistics , and we take advantage of the fact that three instruments recorded energy spectra with good statistics to obtain joint fits . our analysis follows that of frederiks et al .
( 2007a ) , which was based on the konus - wind data .
we also present new , much deeper optical data than previously reported for the section of the refined error ellipse closest to m81 , taken 3 days after grb 051103 .
we use these data to search for possible optical counterparts of this short burst ( sgrb ) , and discuss the implications of its non - detection for its progenitor and putative association with m81 . throughout this paper , we utilize the distance modulus of m81 determined by hst observations of cepheids , 27.8 ( 3.6 mpc , * ? ? ?
grb 051103 occurred at 09:25:42 ut at earth , and was observed by hete - fregate ( atteia et al .
2003 ) , rhessi ( the ramaty high energy solar spectroscopic imager - smith et al .
2002 ) , suzaku - wam ( yamaoka et al . 2009 ) , and swift - bat ( gehrels et al .
2004 ) in low - earth orbit ; the burst was outside the coded fields of view of swift - bat and the hete - ii imaging instruments , and was therefore not localized by them .
it was also observed by integral spi - acs ( rau et al . 2005 ) at 0.5 light - seconds from earth , konus - wind ( aptekar et al . 1995 ) at 4.5 light - seconds from earth , and by mars odyssey ( hend and grs experiments , hurley et al .
2006 ) at 232 light - seconds from earth .
a preliminary ipn error box was announced in a gcn circular , where it was pointed out that this event had the largest peak flux ever observed by konus - wind for a short burst ( golenetskii et al . 2005 ) .
optical follow - up observations were reported by lipunov et al .
( 2005a , b ) , klose et al .
( 2005 ) , and ofek et al .
( 2005 , 2006 ) , and a radio observation was reported by cameron and frail ( 2005 ) .
all yielded negative results . a milagro gev / tev observation similarly produced only upper limits ( parkinson et al . 2005 ) .
the observations of each statistically independent spacecraft pair can be analyzed to produce an annulus of location , and the annuli can be combined to yield an error ellipse using the method described in hurley et al .
( 2000 ) . in this case
, we have combined the konus - hete , konus - rhessi , konus - integral , konus - swift , and konus - odyssey annuli . under these conditions ( several relatively short baselines and one relatively long one ) ,
the error ellipse has a long major axis , corresponding to the annuli derived from the closer spacecraft pairs , and a short minor axis , corresponding to the annulus from the distant spacecraft pair .
we obtain a 3 @xmath4 error ellipse centered at @xmath5=147.933@xmath6 , @xmath7=+68.589@xmath6 , with major and minor axes 137and 0.96respectively , and area 104 square arcminutes .
the chi - square for the error ellipse center is 0.9 for 3 degrees of freedom ( 5 annuli minus 2 fitted coordinates ) .
the area of the initial error box was 240 square arcminutes .
the initial error box and the final error ellipse are shown in figure 1 .
refined error ellipse for the position of grb 051103 and in blue the fields of the region studied using kpno .
the asterisk indicates the center of the ellipse and the most likely arrival direction of the burst .
approximately 65 square arcminutes of the ellipse are contained within the old error box .
these are imposed upon an image of the area surrounding m81 from the digital sky survey .
the possibility that this burst came from the inner disk of m81 is firmly ruled out . however , the brightest galex uv knots noted by ofek et al .
( 2006 ) are within the ellipse .
lipunov ( 2005b ) noted the presence of two galaxies within the initial error box , pgc 2719634 and pgc 028505 .
the former galaxy lies at the 18% confidence contour of the ellipse , and remains a plausible host candidate , while the latter lies at the 0.03% contour , and is unlikely to be the host.,width=309 ]
the rhessi time history of grb 051103 is shown in the top panel of figure 2 .
a distinctive signature of all three previously observed giant sgr flares within our galaxy and the lmc to date is the periodic extended component following the initial short - duration peak . among these three events ,
the periods of this extended tail have clustered around a narrow range of 58 seconds and also have a relatively narrow range of total isotropic energy releases of @xmath8 erg .
this signal lasts for many minutes following the bursts but falls off rapidly after a few hundred seconds . while extended emission is frequently detected following cosmological short - hard bursts , such emission is not periodic
. therefore detection of a periodic component of emission would be considered a strong confirmation of an sgr origin .
none of the ipn light curves shows obvious evidence for extended emission ( pulsed or otherwise ) following the burst .
however , it is conceivable that a marginally detected signal could be present within the noise . to search for such a component , we acquired swift - bat data for grb 051103 ( binned at 64 ms ) and used the lomb ( 1976 ) periodogram to calculate the relative power in the signal following the burst at periods up to about 20 seconds .
we created periodograms for all of the four bat energy channels , which cover the energy range 15 to 350 kev ( and for combinations of channel sums ) and for various time ranges following the emission ( ranging from the first 60 s to the first 300 s. ) to assess the significance of any peaks in the power spectrum , we performed a monte carlo analysis by repeatedly randomizing the order of the 64 ms time bins for each data set over the range of interest and measuring the rate of occurrence of independent peaks above various power levels .
we identified no peaks with greater than 98% significance in any channel or time range .
this non - detection is expected . to assess the general detectability of periodic post - flare emission from extragalactic giant magnetar flares
, we also acquired the swift - bat light curve of the the 2004 december 27th flare from sgr 1806 - 20 .
we then scaled the signal down by a factor of @xmath9 and added it to the light curve of grb 051103 ( both scrambled and unscrambled ) .
no signal is detected in the periodogram at the known periodicity of 7.56 seconds at the distance of m81/82 ( @xmath10 kpc ) .
the maximum distance for detecting periodicity with our analysis greater than 3 sigma is only about @xmath11 kpc ( if a distance of @xmath12 kpc to sgr 1806 - 20 is assumed , or to 130 kpc if the 8.7 kpc distance of bibby et al .
2008 is assumed ) , less than the distance even to m31 .
this limit may not be exact : both sgr1806 - 20 and grb 051103 were detected off - axis by bat and the comparative satellite sensitivity will depend on the specifics of the off - axis angle . for the giant flare from 1806 - 20
, bat was pointing 105away , and slewed to 61away starting around 38 s after the peak .
for 051103 , bat was pointing 122from the source .
however , as the expected signal from a december 27-like event at the distance of m81/m82 would be only 0.01 sigma assuming similar sensitivities for the two events , we consider it extremely unlikely that any possible angle outside the bat fov would lead to a detection unless the periodic component were several orders of magnitude stronger than that observed in the three galactic / lmc events to date .
a key signature of the spectra of the three sgr giant flares observed to date is a very hard energy spectrum for the initial , several hundred millisecond long burst , and a dramatic spectral evolution to a soft spectrum for the subsequent pulsating component .
as these bursts were observed by various instruments , with different temporal resolutions , spectral resolutions , and energy ranges , and all of them were in some degree of saturation at the peak , a precise description of the spectra is impossible .
nevertheless , all of them can be characterized as very hard spectra at the peak , sometimes consistent with a very high temperature blackbody ( e.g. mazets et al .
1979 , fenimore et al . 1981 ,
hurley et al . 1999 ,
mazets et al .
1999 , hurley et al .
2005 , frederiks et al .
accordingly , we have analyzed the time - resolved energy spectra of grb 051103 .
rhessi , konus , and suzaku obtained energy spectra for grb 051103 over a wide energy range , with good statistics , although with different time resolutions .
( due to the off - axis arrival angles at swift and hete - ii , the detector response matrices are not well known , and we have not used these data . ) because the finest time resolution can be obtained from the rhessi data , we have analyzed the rhessi spectra both separately , to obtain the best time resolution , limited only by counting statistics , and combined with the konus and suzaku data , to obtain the best statistics , albeit at the cost of temporal resolution .
rhessi uses nine unshielded coaxial germanium detectors to observe a broad energy band ( 30 kev17 mev ) with excellent energy ( 15 kev ) and time resolution ( 1 binary @xmath13s ) and moderate effective area ( @xmath3150 @xmath14 ) .
the data are recorded event - by - event , which provides great flexibility in choosing analysis intervals . to determine rhessi s spectral response to grb 051103
, we used the monte carlo package mgeant ( sturner et al . 2000 ) .
we simulated monoenergetic photons in 192 logarithmic energy bins ranging from 30 kev30 mev generated along a 60azimuthal arc at the 97off - axis angle of grb 051103 .
we fit a polynomial background and extracted the burst data in solarsoftware - idl routine .
because of radiation damage to some of the detectors , we used only data from rear segments 1 , 4 , 6 , 7 , and 8 .
spectral fitting was conducted with isis v1.4.9 ( houck 2000 ) . in general , the full 30 kev17 mev energy band was employed , except when sufficient counts could not be accumulated at high energies .
we fit the data with a band ( band et al .
1993 ) function : @xmath15 with @xmath16 and @xmath17 . for @xmath18 and @xmath19 , @xmath20 corresponds to the peak of the @xmath21 spectrum . the normalization @xmath22 has units photons/(@xmath14 s kev ) , and @xmath23 is here taken to be 100 kev . for joint fits , the band function parameters @xmath24 , @xmath25 , and e@xmath26 were tied for both instruments , but the normalizations were allowed to vary independently . for the rhessi - only time - resolved fits , we identified time intervals with background - subtracted s / n of 20 in the 60 kev3 mev band .
this yielded three intervals , to which a fourth tail interval of s / n @xmath27 12 was added .
for most intervals , the cutoff power law model ( cpl ) , equivalent to the band function below @xmath28 , provided the best fit .
the time evolution of the parameters of the best - fit spectral model ( a cutoff power - law ) is presented in the lower panels of figure 2 .
the initial spike of emission has a significantly higher peak energy than the decaying tail ; however , the spectral index of the power law appears to harden throughout the burst .
the results are reported in table 1 .
0.04 ms , an e - folding decay time of 28.6 @xmath29 0.6 ms , and a t@xmath30 of 100 @xmath29 4 ms .
the middle and bottom plots show the evolution of the best - fit peak spectral energy and power - law index for the cutoff power - law model .
the black points are rhessi - only , while the gray points are joint fits between rhessi and konus - wind . ]
suzaku - wam did not trigger on grb 051103 , so the only data available are for a 1-second spectrum containing the entire burst , in the 50 kev 5 mev energy range .
the rhessi - wam joint fit is shown in figure 3 and the fit results are reported in table 1 .
the wam fluence is a factor of @xmath35 lower than the rhessi fluence ; this deficit appears to be a result of data lost due to deadtime during the intense peak of emission .
konus - wind triggered on grb 051103 and recorded 64 ms spectra in the 20 kev 10 mev range ; we conducted joint fits between rhessi and konus - wind for the 64 ms and 128 ms intervals analyzed in frederiks et al .
these fits are presented in figures 3 and 4 , and the details are reported in table 1 .
good correspondence was obtained in the best - fit parameters between the two instruments , although a normalization offset was necessary .
[ cols="<,^,^,^,^,^,^,^ " , ] @xmath31 @xcite @xmath32 @xcite @xmath33 @xcite @xmath34 @xcite @xmath35 an approximate fluence calculated using spectral parameters published by @xcite @xmath36 based on observations by @xcite @xmath37 @xcite @xmath38 @xcite @xmath39 @xcite @xmath40 an upper limit based on observations 2 days after the burst completed by @xcite @xmath41 an approximate fluence calculated using spectral parameters published by @xcite @xmath42 @xcite @xmath43 @xcite @xmath44 a lower limit based on observations 4 days after the burst completed by @xcite @xmath45 @xcite @xmath46 a 3@xmath4 upper limit published by @xcite @xmath47 @xcite @xmath48 @xcite the optical afterglows of various sgrbs have been studied and these data can be used to predict the range of afterglow properties of an sgrb of a particular gamma - ray fluence . there is evidence for a reasonable correlation , to first order , between gamma - ray fluence and afterglow flux @xcite . using xspec
, we created a model spectrum of grb 051103 , using the rhessi + kw joint fits in table 1 , and estimated the fluence of grb 051103 in the energy band 15 - 150 kev to be approximately @xmath49erg @xmath50 .
it is possible to compare grb 051103 to other sgrbs in the bat catalogue @xcite using the approximate fluence , calculated for the energy band 15 - 150 kev , and the photon indices given in table 1 .
grb 051103 is isolated at the extreme bright , hard end of the sgrbs in the _ swift _ distribution ( c.f .
figure 14 from * ? ? ?
similarly , in the study of short bursts by mazets et al .
( 2004 ) over the much wider konus energy range , of the 109 spectra which could be characterized by an e@xmath26 , none exceeded 2.53 mev .
the peak energy of grb 051103 is approximately 3 mev ( table 1 ) .
thus if grb 051103 is an sgrb rather than an sgr giant flare , it is a fairly extreme case .
we compared the fluence of this burst to other sgrbs observed by the _
swift _ satellite .
table 3 provides the data of sgrb with fluences in the band 15 - 150kev and late optical observations , obtained from the relevant gcns , and measured optical afterglows in the r band , approximately 3 days after each burst . for two of the bursts it was necessary to estimate the fluence in the correct energy band using the same method as with grb 051103
this is not a complete sample of sgrbs , as there are a number with a relatively low gamma - ray fluence that were either not observed optically , were not observed for longer than a few hours , or did not have a detected optical afterglow .
we chose this sample so we did not have to rely on the assumption that we can extrapolate the light curve to later epochs and because they are of a similar gamma - ray fluence to grb 051103 .
we compared the sgrbs in table 3 to grb 051103 and predict the optical afterglow would have an r band magnitude of @xmath51 as it is at the higher end of the fluence distribution .
this is within the limiting magnitude of the kpno and cfhtls images used , but would have been unobservable in the images obtained by @xcite . as no afterglow was observed , this rules out most typical sgrbs in the region of the error ellipse covered by our imaging .
however , there are cases of sgrbs with extremely faint optical afterglows , for example grb 080503 , which had a similar fluence to grb 051103 and an r band magnitude of [email protected] at 3 days @xcite .
so the observations can not rule out an unusually faint sgrb in this region similar to grb 080503 .
additionally , grb 051103 could be a classical sgrb in the part of the error ellipse not studied in this paper .
conversely , grb 051103 could be an sgr giant flare in m81 with similar energy to the giant flare from sgr 1806 - 20 @xcite and a very faint optical afterglow @xcite .
using observations of the giant flare from sgr 1806 - 20 , we can predict the apparent optical magnitude of an sgr in m81 .
the distance to sgr 1806 - 20 has proven difficult to determine ; the distance modulus adopted by many authors is 15.8 @xcite , although @xcite recently obtained a revised distance modulus estimate of [email protected] .
here we continue to use the larger distance modulus as this will provide an approximate upper limit on the absolute magnitude .
the giant flare from sgr 1806 - 20 had an observed radio afterglow and this has been used by @xcite to make predictions of the apparent r band magnitude of the afterglow .
their analysis suggests that the giant flare would have had an apparent magnitude of @xmath52 at 3 days , and hence an absolute magnitude of @xmath53 . taking this as the absolute magnitude of any afterglow of grb 051103
if it is an sgr giant flare , and using the distance modulus to m81 of 27.8 @xcite , we conclude the afterglow would be expected to have an apparent magnitude of @xmath54 . despite the many uncertainties involved in this calculation , we can have some confidence that such an afterglow would not be detectable with the data available . for future reference ,
it is important to note that with more accurate positions and rapid follow up observations it may be possible to observe the optical afterglows of extragalactic giant flares . for example , if there were a second potential giant flare in m81 we predict the optical afterglow would have a peak apparent k band magnitude of @xmath320 at 86s after the giant flare and would fall to @xmath326 at 1 hour .
this is observable with current and upcoming facilities , for example the european extremely large telescope ( e - elt ) .
however , these predictions are based upon the theoretical models of sgr giant flares being similar to the blast wave model used to describe classical grbs .
@xcite use the blast wave model and radio observations of the giant flare from sgr 1806 - 20 to extrapolate the optical afterglow .
sgrs have been observed during periods of activity using rotse - i @xcite and _ swift _
( for example * ? ? ?
* ) , and ir observations have been obtained for sgr 1900 + 14 4.1 days after outburst detecting no variability @xcite . these have provided upper limits on the optical afterglows from the softer spectrum , shorter , and weaker bursts seen during active phases of sgrs but it is important to note that there have been no reported rapid optical follow up observations of galactic sgr giant flares , which have a significantly higher fluence and are spectrally harder than these bursts .
therefore , we are completely reliant on theoretical predictions and future observations may show discrepancies with these predictions .
indeed , our observations with a limiting magnitude of 24.5 , giving an absolute magnitude -3.3 assuming it is at a distance of 3.6 mpc , constitute one of the deepest absolute magnitude searches for an afterglow from a possible sgr giant flare .
this absolute magnitude is only exceeded by the search for an afterglow from grb 070201 , which is a candidate sgr giant flare in m31 , corresponding to an absolute magnitude of -7.4 obtained 10.6 hours after the burst @xcite .
however , as we discuss later , it is unlikely that both of these events were sgr giant flares @xcite . from the _ galex _ uv imaging @xcite
, there is evidence that the error ellipse does contain star forming regions in the outer disk of m81 .
the two brightest uv sources are marked on figure 6 @xcite .
these young stellar regions in m81 could host an sgr which could emit a giant flare .
similarly , these uv regions could be the locations of massive star clusters , and sgrs 1900 + 14 and 1806 - 20 have been associated with massive star clusters @xcite .
however , if grb 051103 is an sgr giant flare in m81 , we might also expect to find a young ( up to @xmath55 years old , @xcite ) snr in the nearby region , although this association is still being debated @xcite .
when an sgr is formed , it is theoretically possible that it is given a kick of up to @xmath56 km s@xmath57 or more @xcite and therefore could have traveled a distance of @xmath5810pc from the snr .
however , this is only equivalent to an angular separation of @xmath59arcsec at a distance of 3.6mpc @xcite .
hence , an accompanying snr would still be expected to fall within the error ellipse . of the known snr in m81 @xcite , there are none within the error ellipse
. m81 has been studied by the _ chandra x - ray observatory _ @xcite and three x - ray sources are within the error ellipse .
however , they have not been identified in visible or radio observations .
additionally , they have not been identified with known snr , nearby stars , are not co - incident with hii starforming regions ( the expected location of sgrs @xcite ) and are more likely to be x - ray binary systems than unidentified snr @xcite .
this survey had a limiting luminosity of @xmath60 erg s@xmath57 , which means it would detect very young supernovae , as the oldest supernovae with detected x - ray afterglows had a luminosity of @xmath61 erg s@xmath57 and an age of @xmath62 years @xcite .
additionally , this survey would detect the x - ray luminous snr as these have a luminosity of up to @xmath61 erg s@xmath57 but would not detect the x - ray faint snrs which have a luminosity of @xmath63 erg s@xmath57 @xcite .
sgrs are well known to be quiescent soft x - ray emitters and @xcite have measured the soft x - ray flux of sgr 1806 - 20 to be approximately @xmath64erg @xmath50 s@xmath57 .
@xcite determined that the _ chandra observatory _ would be unable to observe directly the persistent x - ray flux from an sgr like sgr 1806 - 20 in m81 .
an alternative method of searching for snr is to use h@xmath24 and oiii narrow band observations .
the isaac newton telescope ( int ) has been used to search for planetary nebulae in m81 by @xcite and they have found 171 potential candidates , some of which are in the nearby region of the refined error ellipse .
their criteria for differentiating between an snr and a planetary nebula is that planetary nebulae can not be spatially resolved and snr are . a young snr , as required for an sgr ,
could be misidentified as a planetary nebula by this criterion , since a one arcsecond region corresponds to a physical size of @xmath65pc .
young snrs may well be significantly smaller than this , since an expansion velocity of a few thousand km s@xmath57 over a magnetar lifetime of @xmath66 years leads to sizes of @xmath67 pc .
indeed , many snrs in m82 appear ( based on radio maps ) to be fairly compact @xcite . however , the nearest is still @xmath323 arcsec from the error ellipse , and we showed earlier that an sgr in m81 would only be able to travel @xmath30.6 arcsec from its birthplace .
the h@xmath24 luminosity of snrs in nearby disk galaxies tends to be greater than @xmath68 erg s@xmath57 @xcite and the work by @xcite had a limiting h@xmath24 flux of less than @xmath69 erg @xmath50 s@xmath57 which corresponds to a limiting luminosity of @xmath70 erg s@xmath57 .
therefore , we would expect their survey to find a candidate snr .
we used the recently published h@xmath24 and oiii images from the int wide field camera imaging survey @xcite , with a limiting luminosity of @xmath71 erg s@xmath57 at the distance of m81 as these are the same images as used by @xcite , and compared them with 21 cm radio images from things @xcite and chandra x - ray source positions @xcite to search for previously unidentified snrs within the error ellipse .
there is a possible circular 21 cm radio source coincident with a chandra x - ray source of approximately the right flux for an snr in m81 ( source 15 in * ? ? ?
* ) . however , the 21 cm radio source is too large for an snr of the required age and there is no convincing supporting evidence of a source within the other images studied .
using the published known x - ray sources we might have expected to find an snr if it was very young or bright and we would have expected to find an associated snr using the h@xmath24 images .
we identified no convincing associated snr candidates within the error ellipse . although it has been determined that the error ellipse does cross potential star formation regions as required by the majority of sgr models , it
should also be noted that this is not essential for all .
an alternative route has been proposed for producing a magnetar by white dwarf ( wd ) mergers @xcite .
as wd have long lifetimes , wd - wd mergers would be associated with older populations of stars .
it is possible that accretion induced collapse ( aic ) will drive off a fraction of the envelope , leaving something akin to an snr behind ( e.g. * ? ? ?
the mechanisms underlying aic are poorly understood , and the physical characteristics and detectability of such remnants are not clear . therefore a sgr produced through this channel could be formed in an old stellar population within the outer disk or halo , and the non - detection of an snr within the region does not place constraints on this model .
if the progenitor was an sgr giant flare , then there might be significant similarities in the light curve and spectrum of grb 051103 to the giant flare from sgr 1806 - 20 .
@xcite noted that the light curve of these two events were consistent , i.e. the light curve of grb 051103 is similar to what would be expected from an extragalactic version of the giant flare from sgr 1806 - 20 . in table 1 , we have shown , for the joint rhessi + kw fits , that initially @xmath72 and softens to @xmath73 .
although this is unusually hard for a grb , it is consistent with the photon index of the giant flare from sgr 1806 - 20 , @xmath74 @xcite .
the peak luminosity of grb 051103 , assuming it was from an sgr in m81 , is approximately @xmath75 erg s @xmath57 .
this is a factor of 10 brighter than the peak luminosity of the giant flare from sgr 1806 - 20 , which is @xmath76 erg s@xmath57 assuming it is at a distance of 15kpc @xcite . with the revised distance estimate from @xcite ,
the peak luminosity of the giant flare from sgr 1806 - 20 would be @xmath77 erg s @xmath57 , suggesting that a much smaller percentage of sgrbs are sgr giant flares .
this value is 30 times fainter than the peak luminosity of grb 051103 if it was from an sgr giant flare in m81 and in this case grb 051103 would be the most luminous sgr giant flare observed . in comparison ,
the peak luminosity of grb 070201 is @xmath78 erg s @xmath57 assuming it was in m31 @xcite , which is an order of magnitude fainter than grb 051103 and comparable to the giant flare from sgr 1806 - 20 .
it is important to note however , that there is currently no theoretical upper limit for the energy of a giant flare .
@xcite showed that the total energy available is given by @xmath79 where @xmath80 .
therefore , the magnetic dipole ( b ) of sgr 1806 - 20 would only need to increase by a factor of @xmath35 to produce a giant flare with an energy that is 30 times greater than the one from sgr 1806 - 20 .
although the gamma - ray data suggest that grb 051103 may be an extragalactic sgr giant flare , it is important to note that sgr giant flares are rare events . considering plausible luminosity functions , @xcite calculated the probability that the ipn would observe a giant flare , with energy greater than the energy emitted by the giant flare from sgr 1806 - 20 , in the region surveyed during the 17 years it has operated . for one giant flare , they calculated the probability to be 10% .
however , as we discussed in the introduction , there are four potential candidates for extragalactic sgr giant flares , including grb 070201 near m31 which has been identified as an sgr giant flare by @xcite .
the probablility that the ipn has detected two sgr giant flares , with energy greater than the giant flare from sgr 1806 - 20 , is 0.6% @xcite .
recently , several new sgr candidates have been identified including 0501 + 4516 , 1550 - 5418 and possibly 0623 - 0006 @xcite , which may imply that the number of sgrs in the milky way is higher than previously thought . in this case , the luminosity of the giant flare from sgr 1806 - 20 would have to be at the peak of the luminosity function of sgr giant flares and therefore giant flares of this luminosity must be extremely rare events .
this argues that grb 051103 is unlikely to be a second sgr giant flare in the nearby universe .
grb 051103 illustrates the difficulties of identifying a short burst as an extragalactic giant magnetar flare beyond a reasonable doubt . even setting aside the questions of detecting and localizing such events , and establishing their associations with nearby galaxies ,
their interpretation is problematic . on the one hand ,
the localization , short duration , and hard energy spectra of grb 051103 suggest that it is a giant flare from m81 .
however , a deeper analysis of its time history demonstrates that the periodic component , which is a key signature of giant flares , is unlikely to ever be detected at great distances by the ipn if all giant flares are similar to the three observed to date .
the energy spectrum at the peak of the emission is very hard ( e@xmath813 mev ) , and is detected to 7 mev at the 3@xmath4 level , with marginal emission up to 17 mev .
yet it is not inconceivable that a short duration grb could have these properties .
although the e@xmath26 of grb 051103 evolves from hard to soft , the evolution to a very soft spectrum , which is expected during the oscillatory phase of an sgr giant flare , is undetectable , as is the oscillatory phase itself .
thus evidence for an extragalactic giant flare origin of grb 051103 remains tantalizing , but inconclusive . on a more positive note ,
if an extragalactic magnetar flare occured within the swift - bat field of view , so that the xrt could begin observing within a minute or so , the periodic component would be detectable at low energies to at least 10 mpc ( hurley et al . 2005 ) .
we have presented new optical observations of grb 051103 and have determined that there is no r band optical afterglow with a limiting magnitude of @xmath82 ( for an afterglow overlapping a host galaxy , the limiting magnitude is @xmath83 ) in the region of the error ellipse covered by our observations .
comparison of the prompt emission of grb 051103 with a sample of other sgrbs leads us to conclude that if it was a classical sgrb we would expect to have located an optical afterglow in our observations .
in contrast , if grb 051103 were an sgr giant flare in m81 , non - detection of an afterglow would not be surprising as the expectations for optical afterglow emission lie significantly below the limits obtained here , or the limits likely to be attained via current technology .
the case for an sgr origin would be strengthened if there were an accompanying snr within the error ellipse , but there is no evidence of this .
an sgr produced via accretion induced collapse of a wd @xcite would , however , remove the requirement for an snr .
additionally , the luminosity of grb 051103 , assuming it is from an sgr giant flare in m81 , is significantly higher than known sgr giant flares but still attainable with current theoretical models .
giant flares with luminosity similar to the giant flare from sgr 1806 - 20 are extremely rare and it is unlikely that grb 051103 and grb 070201 are both extragalactic sgr giant flares .
although we have not considered this option in detail , it is possible that the progenitor of grb 051103 was a compact binary merger in m81 . in this case
, it would just be within the reach of current gravitational wave searches .
this scenario was ruled out at @xmath5899% confidence for grb 070201 in m31 using the laser interferometer gravitational - wave observatory ( ligo ) observations , and distances out to 3.5 mpc were ruled out to 90% confidence @xcite .
the ligo scientific collaboration is currently considering a search for gravitational - wave signals in the data surrounding grb 051103 ( g. jones and p. sutton , private communication ) .
kh is grateful for ipn support from the nasa guest investigator programs for swift ( nasa nng04gq84 g ) , integral ( nag5 - 12706 ) , and suzaku ( nnx06ai36 g ) ; for support under the mars odyssey participating scientist program , jpl contract 1282046 ; and under the hete - ii co - investigator program , mit contract sc - a-293291 .
we are also grateful to the konus - wind team - e. mazets , s. golenetskii , d. frederiks , v. palshin , and r. aptekar - for contributing the konus data to this study .
this research has made use of data obtained using , or software provided by , the uk s astrogrid virtual observatory project , which is funded by the science and technology facilities council and through the eu s framework 6 programme .
the national optical astronomy observatory ( noao ) consists of kitt peak national observatory near tucson , arizona , cerro tololo inter - american observatory near la serena , chilie , and the noao gemini science centre .
noao is operated by the association of the universities for research in astronomy ( aura ) under a cooperative agreement with the national science foundation . based on observations with megaprime / megacam , a joint project of cfht and cea / dapnia , at the canada - france - hawaii telescope ( cfht ) which is operated by the national research council ( nrc ) of canada , the institut national des sceicne de lunivers of the centre national de la recherche scientifique ( cnrs ) of france , and the university of hawaii .
this work is based in part on data products produced at terapix and the canadian astronomy data centre as part of the canada - france - hawaii telescope legacy survey , a collaborative project of nrc and cnrs .
the digitized sky surveys were produced at the space telescope science institute under u.s .
government grant nag w-2166 .
the images of these surveys are based on photographic data obtained using the oschin schmidt telescope on palomar mountain and the uk schmidt telescope .
the plates were processed into the present compressed digital form with the permission of these institutions . based on observations made through the isaac newton group s wide field camera survey programme with the isaac newton telescope operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
s. j. sturner , h. seifert , c. shrader , and b. j. teegarden , 2000 , in american institute of physics conference series , edited by m. l. mcconnell , and j. m. ryan , vol .
510 of american institute of physics conference series , p. 814 . | grb 051103 is considered to be a candidate soft gamma repeater ( sgr ) extragalactic giant magnetar flare by virtue of its proximity on the sky to m81/m82 , as well as its time history , localization , and energy spectrum .
we have derived a refined interplanetary network localization for this burst which reduces the size of the error box by over a factor of two .
we examine its time history for evidence of a periodic component , which would be one signature of an sgr giant flare , and conclude that this component is neither detected nor detectable under reasonable assumptions .
we analyze the time - resolved energy spectra of this event with improved time- and energy resolution , and conclude that although the spectrum is very hard , its temporal evolution at late times can not be determined , which further complicates the giant flare association .
we also present new optical observations reaching limiting magnitudes of @xmath0 , about 4 magnitudes deeper than previously reported . in tandem with serendipitous observations of m81 taken immediately before and one month after the burst , these place strong constraints on any rapidly variable sources in the region of the refined error ellipse proximate to m81 .
we do not find any convincing afterglow candidates from either background galaxies or sources in m81 , although within the refined error region we do locate two uv bright star forming regions which may host sgrs .
a supernova remnant ( snr ) within the error ellipse could provide further support for an sgr giant flare association , but we were unable to identify any snr within the error ellipse . these data still do not allow strong constraints on the nature of the grb 051103 progenitor , and suggest that candidate extragalactic sgr giant flares will be difficult , although not impossible , to confirm .
[ firstpage ] gamma - rays : bursts stars : neutron . |
the milky way halo has long been considered a good target for searches of indirect signatures of dark matter ( dm ) .
wimp dm candidates are expected to produce gamma rays , electrons and protons in their annihilation and decays and such emission originating in our galaxy would appear as a diffuse signal . at the same time , the majority of the galactic diffuse emission is produced through radiative losses of cosmic - ray ( cr ) electrons and nucleons in the interstellar medium .
modeling of this emission presents one of the major challenges when looking for subdominant signals from dark matter . in this analysis
we test the diffuse lat data for a contribution from the dm signal by performing a fit of the spectral and spatial distributions of the expected photons at intermediate galactic latitudes . in doing so , we take into account the most up - to - date modeling of the established astrophysical signal , adapting it to the problem in question @xcite .
our aim is to constrain the dm properties and treat the parameters of the astrophysical diffuse gamma - ray background as nuisance parameters . besides this approach
, we will also quote conservative upper limits using the data only ( i.e. without performing any modeling of the astrophysical background ) .
we follow @xcite in using the ` galprop ` code v54 @xcite , to calculate the propagation and distribution of crs in the galaxy and the whole sky diffuse emission . in @xcite
various standard parameters of the cr propagation were studied in a fit to cr data and it was shown that they represent well the gamma - ray sky , although various residuals ( at a @xmath0 level @xcite ) , both at small and large scales , remain . in our work , we use the results of the fits to the cr data from @xcite but we allow for more freedom in certain parameters governing the cr distribution and known astrophysical diffuse emission and constrain these parameters by fitting the models to _ the lat gamma - ray data_. despite the large freedom we leave in the models we see residuals in our roi at the @xmath1 level and at @xmath2 significance . these residuals can be ascribed to various limitations of the models : imperfections in the modeling of gas and isrf components , simplified assumptions in the propagation set - up , unresolved point sources , and large scale structures like loop i @xcite or the galactic bubbles @xcite . since residuals do not seem obviously related to dm , we focus in the following on setting limits on the possible dm signal , rather than _ searching _ for a dm signal .
we parametrize the smooth dm density @xmath3 with a nfw spatial profile @xcite @xmath4 and a cored ( isothermal - sphere ) profile @xcite : @xmath5 for the local density of dm we take the value of @xmath6 gev @xmath7 @xcite , and the scale radius of @xmath8 20 kpc ( for nfw ) and @xmath9 2.8 kpc ( isothermal profile ) .
we also set the distance of the solar system from the center of the galaxy to the value @xmath10 8.5 kpc .
for the annihilation / decay spectra we consider three channels with distinctly different signatures : annihilation / decay into the @xmath11 channel , into @xmath12 , and into @xmath13 . in the first case
gamma rays are produced through hadronization of annihilation products and subsequent pion decay .
the resulting spectra are similar for all channels in which dm produces heavy quarks and gauge bosons and this channel is therefore representative for a large set of particle physics models .
the choice of leptonic channels provided by the second and third scenarios , is motivated by the dark matter interpretation @xcite of the pamela positron fraction @xcite and the _ fermi _ lat electrons plus positrons @xcite measurements . in this case
, gamma rays are dominantly produced through radiative processes of electrons , as well as through the final state radiation ( fsr ) .
we produce the dm maps with a version of ` galprop ` slightly modified to implement custom dm profiles and injection spectra ( which are calculated by using the ` pppc4dmid ` tool described in @xcite ) .
we use 24 months of lat data in the energy range between 1 and 100 gev ( but , we use energies up to 400 gev when deriving dm limits with no assumption on the astrophysical background ) .
we use only events classified as gamma rays in the p7clean event selection and the corresponding ` p7clean_v6 ` instrument response functions ( irfs ) .
structures like loop i and the galactic bubbles appear mainly at high galactic latitudes and to limit their effects on the fitting we will consider a roi in galactic latitude , @xmath14 , of @xmath15 , and galactic longitude , @xmath16 , @xmath17 .
we mask the region @xmath18 along the galactic plane , in order to reduce the uncertainty due to the modeling of the astrophysical and dm emission profiles . to set these type of limits we first convolve a given dm model with the fermi irfs to obtain the counts expected from dm annihilation .
the expected counts are then compared with the observed counts in our roi and the upper limit is set to the _ minimum _ dm normalization which gives counts in excess of the observed ones in at least one bin , i.e. we set @xmath19 upper limits given by the requirement @xmath20 , where @xmath21 is the expected number of counts from dm in the bin @xmath22 and @xmath23 the actual observed number of counts . in this analysis we model the diffuse emission as a combination of a dark matter and a parameterized conventional astrophysical signal and
we derive the limits on the dm contribution using _ the profile likelihood method_. more precisely , for each dm channel and mass the model which describes the lat data best is the one which maximizes the likelihood function defined as a product running over all spatial and spectral bins @xmath22 , @xmath24 where @xmath25 is the poisson distribution for observing @xmath23 events in bin @xmath22 given an expectation value that depends on the parameter set ( @xmath26 , @xmath27 ) .
@xmath26 is the intensity of the dm component , @xmath27 represents the set of parameters which enter the astrophysical diffuse emission model as linear pre - factors to the individual model components ( cf . equation [ eq : f ] below ) , while @xmath28 denotes the set of parameters which enter in a non - linear way .
we sample non - linear parameters of the astrophysical background on a grid ( for computational efficiency ) .
the linear part of the fit is performed with the ` gardian ` code in the following way .
the cr source distribution ( crsds ) is a critical parameter for dm searches and we define a parametric crsd as sum of _ step functions _ in galactocentric radius @xmath29 , and treat the normalization of each step as a free parameter _ in the fit to gamma rays_. in order to have conservative and robust limits we set to zero the @xmath30 crsds in the inner galaxy region , within 3 kpc of the galactic center . in this way
, potential @xmath31 and @xmath32 cr sources which would be required in the inner galaxy will be potentially compensated by dm , producing conservative constraints .
the linear part of the fit is then performed by combining the sky maps produced in each ring of crsd of the main components of the diffuse emission : the emission from @xmath33 decay , bremsstrahlung and inverse compton .
additionally an isotropic component arising from the extragalactic gamma - ray background and misclassified charged particles needs to be included to fit the _ fermi _ lat data . in summary ,
the various ` galprop ` outputs are combined as : @xmath34 the sum over @xmath22 is the sum over all step - like crsd functions , the sum over @xmath35 corresponds to the sum over all galactocentric annuli ( details of the procedure of a placement of the gas in galactocentric annuli and their boundaries are given in @xcite ) .
@xmath36 denotes the gamma - ray emission from atomic and ionized interstellar gas while @xmath37 the one from molecular hydrogen along the plane , the expression actually simplifies considerably since only the local ring @xmath38 factor enters the sum , since all the other @xmath37 rings do not extend further than 5 degrees from the plane . ] and @xmath39 the inverse compton emission . @xmath40 and @xmath41 are the prompt and inverse compton ( when present ) dm contribution and @xmath42 the overall dm normalization . @xmath43 and @xmath44 denote the isotropic gamma - ray background ( igb ) intensity for each of the five energy bins over which the index @xmath45 runs . in all the rest of the expression
the energy index @xmath45 is implicit since we do nt allow for the freedom of varying the ` galprop ` output from energy bin to energy bin .
we do not include sources in the fit as we use a mask to filter point sources from the 1fgl catalog @xcite .
the outlined procedure is then repeated for each set of values of the non - linear propagation and injection parameters to obtain the full set of profile likelihood curves .
we scan over the three parameters : electron injection index , the height of the diffusive halo and the gas to dust ratio which parametrizes different gas column densities ( see table [ tab : summarypar ] ) . in this way we end up with a set of @xmath28 profiles of likelihood @xmath46 , one for each combination of the non - linear parameters .
the envelope of these curves then approximates the _ final _ profile likelihood curve , @xmath47 , where all the parameters , linear and non - linear have been included in the profile .
limits are calculated from the profile likelihood function by finding the @xmath48 values for which @xmath49 is @xmath50 and @xmath51 , for @xmath52 and 5 @xmath53 c.l .
limits , respectively .
an important point to note is that , for each dm model , the global minimum we found lies within the 3(5 ) @xmath53 regions of many different models , providing a check against a bias in our procedure .
this point is illustrated in figure [ fig : profiles ] , where the profile likelihoods for the three nonlinear parameters , @xmath54 , @xmath55 and d2hi , are shown . to ease reading of the figure
the profiling is actually performed with further grouping dm models with different dm masses , but keeping the different dm channels , dm profiles and the annihilation / decay cases separately .
the curve for the fit without dm is also shown for comparison .
each resulting curve has been further rescaled to a common minimum , since we are interested in showing that several models are within @xmath56 around the minimum for each dm fit .
the @xmath55 profile , for example , indicates that all models with @xmath55 from 1.9 to 2.4 are within @xmath56 around the minimum illustrating that the sampling around each of the minima for the six dm models is dense .
similarly , the d2hi profile indicates that all models with d2hi in the range ( 0.120 - 0.160 ) @xmath57 mag @xmath58 are within @xmath59 from the minima for each of the six dm models .
finally the @xmath54 profile indicates that basically all the considered values of @xmath54 are close to the absolute minima .
this last result is not surprising since , within our low - latitude roi , we have little sensitivity to different @xmath54 and basically all of them fit equally well . there is some tendency to favor higher values of @xmath54 when dm is not included in the fit , while with dm the trend is inverted .
although the feature is not extremely significant it is potentially very interesting .
upper limits on the velocity averaged annihilation cross section into various channels are shown in fig .
[ fig : fixedsourcelimits ] , for isothermal profile of the dm halo , together with regions of parameter space which provide a good fit to pamela ( purple ) and _ fermi _ lat ( blue ) cr electron and positron data @xcite . despite the various conservative choices described above , the resulting dm limits are quite stringent .
they are comparable with the limits from lat searches for a signal from dm annihilation / decay in dwarf galaxies @xcite .
in particular , as shown in figure [ fig : fixedsourcelimits ] for masses around 20 gev the thermal relic value of the annihilation cross section is reached , both for the @xmath60 and @xmath13 channels . overall , rather than being due to residual astrophysical model uncertainties , the remaining major uncertainties in the dm constraints from the halo region come from the modeling of the dm signal itself .
the main uncertainty is in the normalization of the dm profile , which is fixed through the local value of the dm density .
we use the recent determination @xmath6 gev @xmath7 from @xcite , which has , however , a large uncertainty , with values in the range 0.2 - 0.7 gev @xmath7 still viable .
a large uncertainty in @xmath61 is particular important for annihilation constraints since they scale like @xmath62 , while for constraints on decaying dm the scaling is only linear .
a less important role is played by the uncertainties in the dm profile , since in our region of interest different profiles predict similar dm densities . a better determination of the local dm density , as well as of the parameters determining the global structure of the dm halo , is therefore of the utmost importance for reducing the uncertainties related to dm constraints from dm halo , but it is beyond the scope of this paper and is the subject of dedicated studies .
a. w. strong , i. v. moskalenko and o. reimer , astrophys .
j. * 537 * , 763 ( 2000 ) , url : _ http://galprop.stanford.edu / webrun.php_. j. -m .
casandjian , i. grenier [ the fermi - lat collaboration ] , [ arxiv:0912.3478 [ astro-ph.he ] ] .
m. su , t. r. slatyer , d. p. finkbeiner , astrophys .
j. * 724 * , 1044 - 1082 ( 2010 ) .
j. f. navarro , c. s. frenk and s. d. m. white , astrophys . j. * 462 * , 563 ( 1996 )
. k. g. begeman , a. h. broeils and r. h. sanders , mon . not .
soc . * 249 * , 523 ( 1991 ) .
p. salucci , f. nesti , g. gentile , c. f. martins , astron .
astrophys .
* 523 * , a83 ( 2010 ) .
d. grasso _ et al .
_ [ fermi - lat collaboration ] , astropart . phys . * 32 * , 140 ( 2009 ) .
p. meade , m. papucci , a. strumia and t. volansky , nucl .
b * 831 * , 178 ( 2010 ) .
l. bergstrom , j. edsjo and g. zaharijas , phys .
lett . * 103 * , 031103 ( 2009 ) .
o. adriani _ et al . _
[ pamela collaboration ] , nature * 458 * , 607 ( 2009 ) .
a. a. abdo _ et al .
_ [ the fermi lat collaboration ] , phys .
lett . *
102 * , 181101 ( 2009 ) .
m. cirelli _ et al .
_ , jcap * 1103 * ( 2011 ) 051 .
a. a. abdo _ et al .
_ [ the fermi - lat collaboration ] , astrophys . j. suppl .
* 188 * ( 2010 ) 405 - 436 .
.summary table of the parameters varied in the fit .
the top part of the table show the non linear parameters and the grid values at which the likelihood is computed .
the bottom part show the linear parameters and the range of variation allowed in the fit .
the coefficients of the crsds are forced to be positive , apart @xmath63 and @xmath64 which are set to zero .
the local @xmath65 ratio is restricted to vary in the range 0 - 30 @xmath66 @xmath67 ( k km s@xmath68)@xmath68 , while @xmath43 and @xmath69 are left free to assume both positive and negative values .
see the text for more details . [ cols="^,^,^",options="header " , ] @xmath72{mu_an_iso_nosrc_enbin_marcosregions_mycolors_v2_regions2down.pdf } & \includegraphics[width=0.5\columnwidth]{mu_dec_iso_nosrc_enbin_marcosregions_mycolors_v2_rescmdm_regions141up.pdf } \end{array}$ ] @xmath72{tau_an_iso_nosrc_enbin_marcosregions_mycolors_v2_regions2down.pdf } & \includegraphics[width=0.5\columnwidth]{tau_dec_iso_nosrc_enbin_marcosregions_mycolors_v2_rescmdm_regions141up.pdf } \end{array}$ ] | we study diffuse gamma - ray emission at intermediate galactic latitudes measured by the fermi large area telescope with the aim of searching for a signal from dark matter annihilation or decay . in the absence of a robust dark matter signal
, we set conservative dark matter limits requiring that the dark matter signal does not exceed the observed diffuse gamma - ray emission .
a second set of more stringent limits is derived based on modeling the foreground astrophysical diffuse emission .
uncertainties in the height of the diffusive cosmic - ray halo , the distribution of the cosmic - ray sources in the galaxy , the cosmic - ray electron index of the injection spectrum and the column density of the interstellar gas are taken into account using a profile likelihood formalism , while the parameters governing the cosmic - ray propagation have been derived from fits to local cosmic - ray data .
the resulting limits impact the range of particle masses over which dark matter thermal production in the early universe is possible , and challenge the interpretation of the pamela / fermi - lat cosmic ray anomalies as annihilation of dark matter . |
ternary compounds whose structures include a regular sublattice of magnetic moments are attractive objects for studying the influence of magnetism on superconductivity . among these materials ,
the most famous are pbmo@xmath18s@xmath19-type `` chevrel phases '' and ternary rare - earth rhodium borides @xcite .
the physical properties of quadruple compounds dy@xmath20y@xmath21rh@xmath6b@xmath6 having a body - centered tetragonal luru@xmath6b@xmath6-type crystal structure deserve special attention due to a great number of interesting features of these materials . for instance
, it was found @xcite that the magnetic ordering of dy atoms may occur at the temperature @xmath22 higher than the superconducting transition temperature @xmath23 and coexist with superconductivity down to very low temperatures .
it was established in @xcite that dy@xmath20y@xmath21rh@xmath6b@xmath6 belongs to materials with intrinsic ferrimagnetism , and the transition temperature @xmath22 strongly depends on the concentration of non - magnetic yttrium : it falls with increasing y concentration from 37 k in dyrh@xmath6b@xmath6 down to 7 k in dy@xmath24y@xmath25rh@xmath6b@xmath6 . on the contrary
, @xmath23 increases with the y concentration from 4.7 k for dyrh@xmath6b@xmath6 to 10.5 k in yrh@xmath6b@xmath6 @xcite .
measurements of the specific heat of dy@xmath25y@xmath24rh@xmath6b@xmath6 ( @xmath26 k , @xmath27 k ) indicate emergence of another magnetic phase transition below 2.7 k @xcite .
recently , anomalies of some physical quantities , unusual for systems with conventional superconductivity , were discovered in dy@xmath20y@xmath21rh@xmath6b@xmath6 : the paramagnetic meissner effect @xcite and non - monotonic temperature dependencies of the upper critical magnetic field @xmath1 and the superconducting gap @xcite .
another specific feature of this class of magnetic superconductors is the change of the type of magnetic interactions in the dy subsystem under partial replacement of rhodium by ruthenium .
as shown in @xcite , antiferromagnetic ordering in dy(rh@xmath28ru@xmath29)@xmath6b@xmath6 holds for @xmath30 and changes to a ferromagnetic one for @xmath31 .
the superconducting transition temperature also varies with the ru concentration @xcite .
thus , the study of physical properties of the borides family dy@xmath20y@xmath21(rh , ru)@xmath6b@xmath6 with various content of dysprosium ( responsible for the magnetic interactions ) and of ruthenium and rhodium ( responsible for both the magnetic interactions and superconductivity ) is of great interest to explore the coexistence of superconductivity and magnetism and the possibility of appearance of unconventional superconductivity . for this purpose
, we studied in this paper the behavior of the electric resistance in the vicinity of the superconducting transition and the upper critical field in the compound dy@xmath2y@xmath3rh@xmath4ru@xmath5b@xmath6 at different orientations of external magnetic field with respect to the direction of measuring electric current .
the dy@xmath2y@xmath3rh@xmath4ru@xmath5b@xmath6 compound has been prepared by the argon arc melting of initial components , followed by annealing within a few days .
the resulting single - phase polycrystalline ingot had a luru@xmath6b@xmath6-type crystal structure ( space group i4/mmm ) testified by the x - ray phase and structural analyses . at this concentration of ruthenium , it is possible to synthesize samples with such structure at the normal pressure , in contrast to the quadruple compounds dy@xmath20y@xmath21rh@xmath6b@xmath6 , which gain the required structure only during the synthesis under a pressure of 8 gpa .
the samples were cut from the ingot in the form of parallelepipeds whose lengths were about 6 mm and the cross - sectional area was @xmath32 mm .
the measurements of the electrical resistance @xmath8 of the samples were performed on a quantum design ppms-9 automatic system using a standard four - probe circuit with an alternating current ( @xmath33 ma , @xmath34 hz ) in the temperature range @xmath35 k and magnetic fields up to 36 koe produced by a superconducting solenoid .
the sample holder was equipped with a system for automatic rotation of the substrate with the sample by a stepping motor of high resolution which allowed the measurement of @xmath8 for different angles @xmath7 of inclination of the external magnetic field @xmath36 to the direction of the current .
the superconducting transition temperature measured in the middle of the resistive transition in zero magnetic field was @xmath37 k. fig.[fig1 ] presents the temperature dependencies of the sample resistance in different magnetic fields of three orientations : @xmath38 , @xmath39 and @xmath12 ( panels ( a ) , ( b ) and ( c ) , respectively ) .
for the angles @xmath11 and @xmath39 , the experiments were made in magnetic fields @xmath40 koe and for @xmath41 at @xmath42 koe .
the shape of @xmath8 in the range of fields @xmath43 koe is typical for the superconducting transition : a sharp fall of the resistance below a certain temperature followed by its disappearance at lower temperatures .
another type of the behavior of @xmath8 was observed at @xmath14 in the fields larger than 6 koe . in this case , the resistance decreases only down to a certain finite value @xmath44 ( @xmath45 is the sample resistance of in the normal state ) , and then , as the temperature is lowered further , @xmath8 rapidly increases . with further increase of the field , the observed minimum of @xmath8 shifts to lower temperatures and reduces in its depth up to @xmath46 . thus it can be argued that the destruction of superconductivity at the magnetic field orientation @xmath14 ( fig.[fig1]b ) begins in the fields much smaller than at the orientations @xmath38 and @xmath12 .
such strong anisotropy of @xmath8 is apparently due to the presence of the texture ( preferred orientation of individual crystallites ) in a polycrystalline sample and the coexistence of the magnetic and superconducting orderings . along this line of reasonings
, the minimum in @xmath8 for @xmath14 can be attributed to stronger ( compared to other orientations ) enhancement of an uncompensated magnetic moment with increasing magnetic field .
this excess magnetism leads to significant suppression of the superconducting state in the fields @xmath47 koe at @xmath14 , while for other orientations , superconductivity holds up to several tens of koe .
= 8.5 cm = 8.5 cm we note that a minimum of @xmath8 in magnetic fields has been earlier observed in other magnetic superconductors such as ndrh@xmath6b@xmath6 @xcite and dy@xmath48mo@xmath18s@xmath19 @xcite .
it has been attributed to the induction of the magnetic ordering of nd and dy ions by an external magnetic field at the transition temperature @xmath49 , which leads to destruction of the superconducting state ( reentrant superconductivity ) . however , in our case , the superconductivity and the magnetic order , which emerges at @xmath50 , coexist below @xmath23 , and the minimum in @xmath8 can be explained as the result of changes in the existing magnetic structure caused by the magnetic field of specific orientation . using the data of fig.[fig1 ] and accepting for @xmath9 the values of the external magnetic field and the temperature , at which @xmath51 , we plotted the experimental temperature dependencies of the upper critical field depicted by circles and squares in fig.[fig2 ] .
in contrast to the previous andreev spectroscopy data @xcite , we did not found any non - monotony in the behavior of @xmath9 that possibly reflects certain ambiguity in the interpretation of the results of the point - contact measurements in nonhomogeneous samples ; a certain role in this difference may be also played by the admixture of ruthenium in our samples . dashed curves in fig.[fig1 ]
show the dependencies @xmath9 calculated from the equation of the werthamer - gelfand - hohenberg ( whh ) theory @xcite : @xmath52 here @xmath53 is the digamma function , @xmath54 , and @xmath55 are the reduced temperature and critical magnetic field , respectively . in our calculations
, we use the fitting values of the maki parameter @xmath10 which describes relative contribution of the spin - paramagnetic effect and the parameter @xmath56 of the spin - orbit scattering . the best fit for @xmath11 and @xmath12
gives @xmath57 , i.e. , only the orbital effect is responsible for the suppression of superconductivity , whereas at @xmath14 we obtain a rather large value of @xmath58 ( @xmath59 ) which indicates essential contribution of the spin - paramagnetic effect .
figure 2 shows that for @xmath60 , the experimental values of @xmath9 at the temperatures below @xmath61 slightly exceed the maximum possible calculated ones .
this could be explained either by the effect of anisotropy ( may lead to increase in @xmath1 by @xmath62 @xcite ) or by the presence of strong coupling in the superconducting condensate ( can enhance @xmath1 by @xmath63 times @xcite ) , which are beyond the frameworks of the whh theory .
the possibility of the strong coupling in this material follows from the results of the andreev spectroscopy of the superconducting gap @xmath64 @xcite , according to which the ratio @xmath65 can reach 4 or even higher values , larger than the value 3.52 for conventional superconductors with the weak coupling .
the orbital critical field at @xmath66 can be calculated by using the formula of the whh theory for the dirty limit @xcite , @xmath67 while the upper critical field can be estimated as @xcite @xmath68 as is obvious from fig.[fig2 ] , initial slopes of @xmath9 near @xmath23 are approximately equal for all orientations of the magnetic field . according to , this results in a universal ( angle - independent ) value of @xmath69 koe . since at @xmath11 and @xmath12
the fitting value of the maki parameter is close to zero , we conclude that the orbital field at these orientations fully determines the magnitude @xmath70 of the upper critical field at zero temperature .
the estimate of a small paramagnetic contribution can be obtained from the relation @xmath71 for the chandrasekhar - clogston limit @xcite ( @xmath72 is the bohr s magneton ) which gives the critical paramagnetic field @xmath73 koe .
then , using the equation @xcite @xmath74 we found a rather small value @xmath75 for the maki parameter which explains unobservability of the spin effects at the experimental dependencies @xmath9 at @xmath11 and @xmath12 .
as noted above , at @xmath14 , the large value of the maki parameter @xmath15 implies that the spin - paramagnetic effect plays the main role in the suppression of superconductivity . in this case
, equation gives @xmath76 koe , i.e. , by 4.3 times smaller than its value for the orientations @xmath11 and @xmath12 .
correspondingly , the magnitude of the critical paramagnetic field @xmath77 koe found from appears to be much smaller than at @xmath11 and @xmath12 .
these results give additional arguments in the benefit of our assumption about rearrangement of the magnetic structure and formation of an excess magnetic moment induced by the external magnetic field .
this considerably enhances the effective magnetic field acting on the electron spins which causes destruction of the cooper pairs .
= 8.5 cm in order to obtain additional information about the effect of the magnetic field inclination on the suppression of superconductivity , we measured the angle dependencies of the sample resistance within the range @xmath78 ( see fig.[fig3 ] ) at the temperature @xmath79 k which corresponds to the middle of the superconducting transition at @xmath80 in the field of 8 koe . figure [ fig3 ] shows that with an increase in the angle , the resistance first grows to a maximum value @xmath45 at @xmath14 , then begins to drop with a minimum at @xmath81 .
all highs in @xmath82 repeat themselves through @xmath12 , and lows through @xmath83 ( the magnitude of minimum ohmic losses for @xmath11 and @xmath83 is smaller than that at @xmath81 and @xmath84 ) .
thus , the superconductivity is most strongly suppressed in the magnetic fields directed at the angles @xmath39 plus multiple of @xmath12 relative to the longitudinal sample axis .
the fields inclined at the angles by multiple of @xmath12 have the weakest impact on the superconducting state .
the dashed line in fig.[fig3 ] indicates the experimental data obtained at @xmath85 k in the field of 8 koe and demonstrates independence of the sample resistance of the field direction in the normal state .
we have measured the resistance @xmath0 and the upper critical magnetic field @xmath1 of the magnetic superconductor dy@xmath86y@xmath3rh@xmath4ru@xmath5b@xmath6 at different angles @xmath7 of inclination of the magnetic field relative to the longitudinal sample axis .
the temperature dependencies @xmath8 and @xmath9 are strongly anisotropic in the superconducting state , whereas the rotation of the magnetic field in the normal state has no effect on its resistive properties . suppression of superconductivity is most pronounced at @xmath87 plus multiple of @xmath12 ( @xmath76 koe ) , while at the angles @xmath60 and @xmath12 the effect of the magnetic field on the superconducting state is weakest , and the calculated magnitude of @xmath16 reaches 38 koe .
a minimum in @xmath8 in large enough fields was observed at @xmath87 which resembles reentrance effects in some magnetic superconductors near the point of transition to the magnetically ordered state .
however , in our case , this minimum most likely occurs due to restructuring of already existed magnetic ordering .
analysis of the behavior of @xmath9 within the framework of the whh theory shows that for @xmath11 and @xmath12 the maki parameter @xmath10 and the parameter @xmath56 of the spin - orbit scattering are close to zero , i.e. , only the orbital effect is responsible for the suppression of superconductivity .
this is confirmed by the estimate of @xmath10 obtained from the calculated paramagnetic limit . on the contrary , at @xmath14 ,
the maki parameter was found to be large ( @xmath58 , @xmath59 ) which manifests the dominating role of the spin - paramagnetic depairing mechanism .
we suggest that the above mentioned features of the behavior of superconducting parameters may be associated with the growth of a spontaneous magnetic moment of the dysprosium subsystem induced by an external magnetic field of specified orientation . at the same time
, one can not exclude the existence of an unconventional pairing mechanism , such as a triplet pairing , in this material .
99 v.m .
dmitriev , a.j .
zaleskii , e.p .
khlybov , l.f .
rybaltchenko , e.v .
khristenko , l.a .
ishchenko , a.v .
terekhov , i.e. kostyleva , s.a .
lachenkov , fizika nizk . temp .
34 ( 2008 ) 1152 [ low temp . phys . 34 ( 2008 )
dmitriev , a.v .
terekhov , a. zaleskii , e.n .
khatsko , p.s .
kalinin , a.i .
rykova , a.m. gurevich , s.a .
glagolev , e.p .
khlybov , i.e. kostyleva , s.a .
lachenkov , fizika nizk . temp .
38 ( 2012 ) 191 [ low temp
38 ( 2012 ) 154 ] . | we have measured temperature dependencies of the electric resistance @xmath0 and upper critical magnetic field @xmath1 of a magnetic superconductor dy@xmath2y@xmath3rh@xmath4ru@xmath5b@xmath6 .
the measurements were made for different angles @xmath7 of magnetic field inclination to the direction of measuring current and revealed strong anisotropy of the behavior of @xmath8 and the values of @xmath9 . by using the werthamer - gelfand - hohenberg theory , we determined the maki parameter @xmath10 and the parameter of the spin - orbital interaction . for @xmath11 and @xmath12
both parameters are close to zero , thus the magnitude of @xmath13 koe is basically limited by the orbital effect . at @xmath14 , a large value of the parameter @xmath15 indicates dominating role of the spin - paramagnetic effect in the suppression of @xmath16 down to @xmath17 koe .
we suggest that such behavior of @xmath8 and @xmath9 is caused by internal magnetism of the dy atoms which may strongly depend on the magnetic field orientation .
magnetic superconductor , rare - earth borides of rhodium , upper critical field 74.25.ha , 74.25.dd , 74.70.fc |
the unprecedented sensitivity and angular resolution of _ chandra _ has been utilized by wang _
et al . _ ( 2002 ; hereafter w02 @xcite ) and muno _ et al . _ ( 2003 ; hereafter m03 @xcite ) to investigate the x - ray source population of the galactic center ( gc ) .
the w02 acis - i survey of the central 0.8@xmath0@xmath12@xmath0of the gc revealed a population of @xmath2800 previously undiscovered discrete weak sources with x - ray luminosities of @xmath3@xmath4erg@xmath4s@xmath5 .
m03 imaged the central 40 pc@xmath6 ( at 8 kpc ) around sgr a * , finding an additional @xmath22300 discrete point sources down to a limiting flux of @xmath7 erg / s .
more recently , a deeper _ chandra _ survey of the central @xmath21@xmath0around sgr a * has been obtained ; the combination of all of these surveys has revealed a total of @xmath810,000 discrete sources ( figure 1 ; muno _ et al . _ 2008 ,
hereafter m08 @xcite ) .
the harder ( @xmath93 kev ) x - ray sources are likely to be at the distance of the gc , while the softer sources are likely to be foreground x - ray active stars or cvs within a few kpc of the sun .
some individual sources have been identified as x - ray transients , high - mass stars , lmxbs , and cvs
. however , the nature of the majority of these newly detected sources is as yet unknown .
we have undertaken an ir and x - ray observational campaign to determine the nature of the faint discrete x - ray source population discovered by _
chandra _ in the gc .
we have begun by cross - correlating the source catalog of muno _ et al . _ ( 2006 ; hereafter m06 @xcite ) with @xmath10images of 26 selected 2.5@xmath11@xmath6 regions obtained with isaac on the vlt to identify candidate ir counterparts to the x - ray sources @xcite . our original search for the counterparts to the gc x - ray sources used positions obtained through private communication with the authors of the w02 survey .
however , m06 re - analyzed the data and produced a significant refinement of the astrometry of the w02 data ; thus the positions reported for @xmath250% of the x - ray sources in the finalized m06 catalog showed a shift of @xmath12@xmath11 from the positions we used for our original cross - correlation analysis .
this astrometric shift has a substantial impact on the identification of candidate ir counterparts , since the average stellar separation in our vlt fields is 1.94@xmath11 in @xmath13-band @xcite .
in addition , the revised m06 analysis revealed that a larger number of x - ray sources were detected in the original _
chandra _ survey than had been previously cataloged . as a result
, our vlt imaging survey covers 130 x - ray sources , compared to the 77 x - ray sources originally reported in @xcite . ; _ middle , green _ : @xmath14 ; and _ right , red _ : @xmath15 .
the large extension towards high values of colour ( _ right _ in each panel ) indicates that there is a significant amount of near - ir extinction.,scaledwidth=99.0% ] comparing the colours and magnitudes of the candidate counterparts to the general field populations ( figure 2 ) , we find no clearly significant difference between the magnitude distribution of candidate counterparts and those of the field population , aside a slight over - abundance of more local , bright , blue candidate counterparts ( see especially the left - hand panel of figure 2 ) .
the colours indicate that the majority of potential counterparts are located at or beyond gc distances ( @xmath16 ) , with a smaller number consistent with a foreground stellar population .
@xmath17 ( 40 ) of the x - ray sources that lie within the area covered by our vlt observations do not have detections within the x - ray error circle in any of the three ir bands .
@xmath18 ( 76 ) of the x - ray sources had zero candidate counterpart detections in the @xmath19-band , compared to @xmath20 ( 63 ) for the @xmath21-band and only @xmath22 ( 47 ) for the @xmath23-band . to examine the significance of these numbers of `` non - matches ''
, we performed a monte carlo test .
we added a random 0@xmath11 30@xmath11 shift to the position of each x - ray source , re - ran the matching of the x - ray and ir data , and recorded the number of candidate counterparts identified for the match ; we performed this test 10,000 times , having different random shifts in the x - ray position for each trial .
the monte carlo test revealed that , given randomized x - ray positions , the expected fraction of x - ray sources that would have no ir counterparts across all three bands is [email protected]% ( @xmath25 error ) . for the actual positions , @xmath17 of x - ray sources had no candidate ir counterpart , which is @xmath26 _
lower_. similarly for @xmath19 , @xmath21 and @xmath23-bands considered individually , the monte carlo simulation revealed that the fraction of x - ray sources for which there would be no matches if the x - ray sources are randomly scattered across the fields are [email protected]% , [email protected]% and [email protected]% respectively .
the numbers of `` non - matches '' for the _ true _ x - ray positions are thus _ lower _ than would be expected from random at the @xmath27 , @xmath28 and @xmath29 levels for @xmath19 , @xmath21 and @xmath23 respectively .
this result indicates that even with the extremely high ir source density in the gc , with the corrected and accurate x - ray astrometry we are now finding significantly _ more _ x - ray / ir matches than would be expected from chance coincidence .
from our analysis of the vlt images , we find that the ir extinction in the gc can vary on scales as small as 5@xmath11 ( 0.2 - 0.6 pc at 8 kpc ; @xcite ) .
some areas show little evidence of this `` granularity '' , while others are highly structured .
the relationship of extinction to wavelength in the ir is a power law with slope @xmath30 @xcite ; the `` canonical '' value for @xmath30 is @xmath22 @xcite . in contrast , for the gc we find a mean value of @xmath31 ; and furthermore , along any given line of sight to the gc @xmath30 varies substantially , ranging from @xmath21.83.6 @xcite .
we find that the `` universal '' ir extinction law is _ not _ universal in the gc . therefore to obtain reddening - corrected stellar photometry in the gc , a local value for the @xmath10extinction ( on scales @xmath32@xmath11 ) must be measured and applied .
the primary accretion signature in the @xmath13-band which distinguishes a true x - ray counterpart from a field star is strong brackett @xmath33 emission ; this technique of identifying xrb counterparts has been verified with observations of several well - studied gc xrbs @xcite . with this technique , using spex on irtf we conclusively identified a heavily - reddened early - type star as the ir counterpart to the gc source `` edd-1 '' ( cxogc j174536.1 - 285638 ; @xcite ) . however , for an additional 16 candidate counterparts for which we obtained @xmath13-band spectra with isaac , we did not detect br-@xmath33emission .
a possible explanation is that the accretion signatures could be too weak to be measureable , for example if the accretion rate was low at the time of observation , or if the emission was self - absorbed by the mass donor .
however , a more likely explanation is that the stars we observed are not the true counterparts to the x - ray sources .
our imaging survey had a limiting magnitude of @xmath13=20 , so our spectroscopic data would only detect xrbs with either early - type ( as in `` edd-1 '' ) or evolved mass donors . therefore it is likely that the majority of the true ir counterparts belong to a lower mass population of stars which includes the mass donors of cvs and lmxbs ( as has been suggested by @xcite , and other authors ) ; at gc distances , these counterparts are fainter than the limits of our survey .
cross - correlation of our ir imaging of the gc with the _ chandra _ surveys will produce a large number of ir candidate counterparts to the x - ray sources .
due to the extremely high stellar density in the nuclear bulge , many of these astrometric `` matches '' are likely to be chance superpositions , as indicated by our initial spectroscopic data . with thousands of candidate counterparts , traditional long - slit single - target spectroscopy would be a prohibitively inefficient method by which to identify true counterparts .
thus we will need to follow - up with multi - object ir spectroscopy to find the true matches : this is the work which will be performed with the flamingos-2 galactic center survey . | i describe the ir and x - ray observational campaign we have undertaken for the purpose of determining the nature of the faint discrete x - ray source population discovered by chandra in the galactic center ( gc ) .
data obtained for this project includes a deep chandra survey of the galactic bulge ; deep , high resolution ir imaging from vlt / isaac , ctio / ispi , and the ukidss galactic plane survey ( gps ) ; and ir spectroscopy from vlt / isaac and irtf / spex . by cross - correlating the gc x - ray imaging from chandra with our ir surveys ,
we identify candidate counterparts to the x - ray sources via astrometry . using a detailed ir extinction map
, we are deriving magnitudes and colors for all the candidates .
having thus established a target list , we will use the multi - object ir spectrograph flamingos-2 on gemini - south to carry out a spectroscopic survey of the candidate counterparts , to search for emission line signatures which are a hallmark of accreting binaries . by determining the nature of these x - ray sources
, this flamingos-2 galactic center survey will have a dramatic impact on our knowledge of the galactic accreting binary population .
address = dept . of astronomy ,
university of florida , gainesville , fl 32611 usa address = dept . of astrophysics ,
university of oxford , oxford ox1 3rh , uk , altaddress = department of physical sciences , p.o .
box 3000 , 90014 university of oulu , finland address = dept . of astronomy ,
university of florida , gainesville , fl 32611 usa address = california institute of technology , pasadena , ca 91125 usa address = dept . of astrophysics ,
university of oxford , oxford ox1 3rh , uk address = dept . of astrophysics ,
university of oxford , oxford ox1 3rh , uk address = dept . of astronomy ,
university of florida , gainesville , fl 32611 usa address = dept . of astronomy ,
university of florida , gainesville , fl 32611 usa |
we consider a spherical in shape colloidal particle with mesoscopic radius @xmath4 , or two such particles with , in general , different mesoscopic radii @xmath5 and @xmath6 immersed in a medium that is either simple fluid or a binary liquid mixture .
we envisage the case in which the particle is placed at a distance @xmath0 of closest surface - to - surface approach from a planar boundary wall , or the situation in which the surfaces of the two spheres are at such a distance @xmath0 from each other .
we are interested in the force @xmath7 between the particle and the plane or into that one between the two particles .
such fluid mediated forces between two surfaces or large particles are usually referred to as solvation forces @xcite in colloid science .
here we are going to consider the special case when the temperature of the medium is close to the critical temperature @xmath8 of either the liquid - vapor critical point of the simple fluid or the critical demixing point of a binary liquid mixture . in this case
the solvation force acquires , as pointed out by fisher and de gennes @xcite , a contribution due to the critical fluctuation of the medium .
this contribution is of a long - ranged character .
it is characterized , to a great extend , by the gross features of the medium @xcite depending also on the boundary conditions which the bodies immersed in the medium impose on it at their surfaces .
this fluctuation induced force , which is due to the critical fluctuations of the order parameter of the medium is , thus , _
universal _ in nature .
it has a lot of similarities with the force between neutral bodies due to the quantum and temperature fluctuations of the charge distributions in them , i.e. , of the electromagnetic field , which force is known today under the general name of a casimir force ( cf ) @xcite , or , more specifically , quantum electrodynamic ( qed ) cf .
that is why the fluctuation part of the solvation force @xmath9 near @xmath8 is termed _
critical casimir force _ ( ccf ) @xcite .
the force @xmath7 separates into a regular background contribution @xmath10 , which depends on the parameters characterizing the medium in an analytic way , and a singular contribution @xmath11 , which is due to the critical fluctuations of the medium @xmath12 obviously , the grand canonical potential of the system @xmath13 , or the free energy , of the system depends on the geometrical characteristics like @xmath0 , @xmath4 , @xmath5 and @xmath6 , where @xmath1 and @xmath2 are the temperature and the chemical potential of the fluid .
then one has @xmath14 one normally defines @xmath15 within the system considered we suppose that the colloid particles involved , the solid wall and the medium are all governed by dispersion , van der waals type , interactions .
thus , one shall have @xmath16 it shall be stressed that both the ccf and van der waals force ( vdwf ) are fluctuation induced ones but due to the fluctuations of different entities
the first is due to massless excitations of the order parameter , while the second is due to that ones of the electromagnetic field . in the current article
we are going to consider how the interplay of these two type of interaction governs the behavior of colloidal particles .
currently there is no general theory available to scope with the problem of quantitative description of the mutual influence of the fluctuation of the electromagnetic field and the order parameter fluctuations of a medium when it is close to its critical point .
the lifshitz theory @xcite which is the basic one for studying the casimir effect due to the fluctuations of the electromagnetic field has never been meant to nor can deal with the problem of a critical medium between two other media . for practical applications of this theory the main quantity ,
which knowledge for any material involved is required , is the dielectric permittivity @xmath17 . under the normal approach
it is usually tabulated at room temperature for specific values of the angular frequency @xmath18 of the electromagnetic field .
then , in order to perform the specific calculations needed one supposes the analytic validity of a given dependence on @xmath18 , say , the validity of the drude or plasma models for the considered material . it shall be emphasized , however , that in a critical fluid @xmath17 is itself a _ singular _ function of the temperature @xcite .
we are not aware of a theory that quantitatively predicts how @xmath17 depends on the temperature and @xmath18 near the critical point of the medium for a specific material characterized by some characteristic spectrum . due to the hypothesis of universality and scaling when studying critical phenomena one normally is interested only in the gross features of the system and utilizes some effective hamiltonian where only few basic features of the critical medium are reflected . in @xcite
an approach has been suggested for such a system in the case of a film geometry , based on the equation of state of the critical medium , which provides a uniform treatment of the contributions due both to the vdwf and the ccf .
it uses general renormalization group arguments and makes use of field - theoretical methods within mean - field approximation .
this relatively simple model allows all the calculations for the forces involved to be performed on equal footing .
let us note that the expression derived there for the hamaker term is in full agreement with the dzyaloshinskii - lifshitz - pitaevskii theory @xcite .
let us also remind that the mean - field theory is considered as a reliable theoretical horse for the qualitative description of the critical phenomena . in the current article
we extend the result of @xcite towards systems with more complex geometry sphere - plane and sphere - sphere .
the interest in the fluctuation - induced phenomena in the last years blossomed due to their importance in the rapidly developing field of nanotechnology where , below a micrometer distances , the vdwf and qed cf play a dominant role between neutral nonmagnetic objects .
the last implies that these forces play a key role in micro- and nano - electromechanical systems ( mems / nems ) @xcite operating at such distances . in vacuum , or gas medium , they lead to irreversible , usually undesirable phenomena , such as stiction ( i.e. , irreversible adhesion ) or pull - in due to mechanical instabilities @xcite . closely related to that is another troubling effect : when a particle s characteristic size is scaled down below a micrometre the role of its weight becomes negligible . as a result ,
when one tries to release such a neutral particle from whatever handling device surface operating on it in air or under vacuum , the particle will not drop down under the gravity but , instead , will stick to its working surface due to the effect of the omnipresent vdwf .
if , in an attempt to release the particle , one charges the particle , forces vibration of the working surface , etc .
, the released particle might move in an uncontrollable way leaving the observation field of the apparatus controlling the performance of the operation .
that is the main reason why the handling , feeding , trapping and fixing of microparts is still the main bottleneck in micro manufacturing and is far from being solved in a satisfactory fashion @xcite .
thus , formalizing the above , one of the main problems in the micro- and nano - assembly is the precise and reliable manipulation of a micro- or nanopart that includes moving it from a given starting point , where it is to be taken from , to some end point , when it is to be placed on . in that respect
it seems ideal , if one can modify the total interaction between the manipulated particle and the operating device , sometimes called gripper , in such a way that it is repulsive for short distances between the working surfaces and the particle and attractive at larger ones .
it is clear that the ability to modify the casimir interaction can strongly influence the development of mems / nems .
several theorems , however , seriously limit the possible search of repulsive qed cf @xcite .
currently , apart from some suggestions for achieving qed casimir repulsion in systems out of equilibrium the only experimentally well verified way to obtain such repulsive force is to have interaction between two different materials characterized by dielectric permittivities @xmath19 and @xmath20 such that @xcite @xmath21 along the imaginary frequency axis , with @xmath22 being the dielectric permittivity of the medium between them . in refs .
@xcite qed casimir repulsion was indeed observed experimentally for the sphere - plate geometry . in order to minimize the potential negative effects of all possible circuitry at such a small distances and the complications with the isolation ,
as well as possible problems involving chemical reactions it seems that one promising strategy for overcoming the obstacles mentioned above is to chose such a fluid as a medium that possesses no free changes dissolved in it and that is inert and do not interact chemically with the materials .
that leads us of choosing as a fluid a nonpolar liquefied noble gas that has critical parameters as close as possible to the normal ones . in the current article
we will demonstrate that by proper choice of the materials of the colloid particles and the working surface of the gripper it is indeed possible to achieve control over the net interaction between the surface and the particle by simply changing the temperature or the chemical potential in the system .
we also present results for the force between two colloid particles .
let us stress that due to its unique temperature dependence , ccf allows _ in situ _ control of reversible assembly in soft matter and nanoscience .
a further advantage of the force is that both its magnitude and range of action depends on @xmath1 and @xmath2 .
we demonstrate this for the case of fluid systems consisting of a nonpolar medium , a solvent , @xmath23 comprising two objects of materials @xmath24 and @xmath25 immersed in the solvent , which are coated by thin solid films of substances @xmath26 and @xmath27 , respectively see fig . [
fig : fluid_systems ] . due to
both scientific and technological reasons currently the casimir effect is object of intense studies .
the last is true both for the qed casimir effect , as well as in its thermodynamic manifestation via the ccf . in the current article
we will report theoretical results referring to the critical casimir effect between pair of objects , at least one of which has spherical shape .
let us stress out that the ccf has already been measured @xcite , utilizing light scattering technique , in the interaction between a single colloidal sphere and a flat silica surface immersed in a binary mixture near its critical point .
the theoretical background of the obtained results is discussed in details in ref .
[ ] . in ref .
[ ] , using a system of three optically trapped @xcite spherical colloidal particles , immersed in a critical binary mixture , the authors demonstrated experimentally the theoretically predicted nonadditivity of the fluctuation - induced interactions .
other experimental setups which exploit a sphere - plate or sphere - sphere configurations immersed in a critical fluid include the interaction of spherical colloids with chemically patterned substrates @xcite , formation of critical colloid aggregates @xcite , phase behaviour studies @xcite , various techniques for fine - tuning of the ccf @xcite .
as far as the theoretical side is concerned , it was de gennes who first obtained the ccf between spherical particles @xcite considering a local free - energy functional . among the other techniques used to study the ccf in sphere - plate and sphere - sphere geometries are the ornstein - zernicke theory @xcite , conformal invariance methods @xcite , monte carlo calculations @xcite , fluid - particle dynamics simulations@xcite , mean - field type @xcite and density - functional @xcite theory calculations combined with the derjaguin approximation @xcite . several review articles and works
@xcite summarize both the experimental and theoretical results presented there .
as already stated above , in the current article we study the interplay between the ccf and the dispersion vdwf arising between a sphere and solid plate and between two spheres systems immersed in a solvent which is close to its critical point .
the solvent is thought to be either a nonpolar fluid or binary liquid mixture consisting of nonpolar components that is close to its demixing point .
the interacting objects are assumed coated by thin solid films .
we suppose that the liquid phase of the one - component fluid or one of the components of the binary liquid mixture are strongly adsorbed by _ both _ coating layers , i.e. , they ensure the so - called @xmath3 boundary conditions . the bulk phase ( core ) of the objects , on the other hand , influence the fluid by long - range _ competing _ dispersion potentials . in the case of a simple
fluid these potentials increase the adsorption of the fluid near one of the surfaces , leading to preference there of its liquid phase , and decrease it near the other one . in the case of a binary fluid mixture
the substrates prefer one of the components near , say , the top and the other one near the bottom of the system .
we will demonstrate that this experimentally realizable competition between the effects due to the coatings and the cores can result in interesting effects like sign change of the ccf , as well as the ntf , acting between the surfaces , when one changes @xmath1 , @xmath2 or @xmath0 .
the last facts can potentially be used in controlling the properties of colloidal suspensions and for governing the behavior of objects at small , below micrometer , distances .
the content of the article is arranged as follows . in sec .
[ sec : general ] we present some general predictions of the finite - size scaling theory for the interaction between sphere and a plate and between two spheres . in subsec .
[ sec : thermalcas ] we recall and comment on the finite - size behaviour of systems with dispersion forces extending the known facts to the expected behavior of the ccf , vdwf and ntf when they act between pair of parallel plates , sphere and a plate and couple of spherical particles . by doing so , we introduce and compare the two general techniques , namely the derjaguin approximation ( da ) ( subsec .
[ sec : da ] ) and the surface integration approach ( sia ) ( subsec .
[ sec : sia ] ) , within which we study the commented spectrum of forces .
section [ sec : model ] briefly presents the corresponding lattice - gas model suitable for the investigation of fluid media with account of the long - ranged van der waals interactions . here
we present the equation for the equilibrium profile of the finite - size order parameter , identify the main coupling parameters characterizing the interactions in the systems , which enter in it , and give the general expressions used to calculate the ccf and ntf .
last but not least in sec .
[ sec : resultsandexpfeas ] we present the exact equations ( subsec .
[ subsec : spplspspdasiad3 ] ) used to obtain the numerical results for the behaviour of the investigated forces , and comment them in details in subsec .
[ subsec : discussion ] .
finally , we provide arguments in support of the experimental feasibility of the predicted effects subsec .
[ subsec : experimentalfeas ] .
the article ends with a summary and discussion section sec .
[ sec : disandconcrem ] .
important technical details concerning the derivation of the expressions for the interaction forces between spherical particles within the sia are presented in appendix [ sec : appspspsia ] .
since the ccf depends on the critical properties near the bulk critical temperature @xmath8 of the solvent it is governed by universality and scaling @xcite . the last implies that , in first approximation , the critical casimir effect depends only on the gross features of the system its dimensionality @xmath28 and the symmetry of the ordered state ( both defining the so - called bulk universality class of the system ) and on the boundary conditions ( determined by the surface universality classes ) imposed on the fluid by the bodies immersed in it .
therefore , to a great extend the ccf is _
universal_. the quantitative effects of the presence of a surfaces of the bodies on the thermodynamic behavior of the system depends on the penetration depth of their symmetry breaking effect into the volume . obviously , the range to which these effects are felt within the system depends on two phenomena : on how long - ranged the interactions are and on how long - ranged the fluid correlations , which mediate the interactions between the bodies , are .
the long - ranginess of the correlations is set by the correlation length @xmath29 of the order parameter of the solvent ; @xmath29 becomes large , and theoretically diverges , in the vicinity of the bulk critical point @xmath30 : @xmath31 , where @xmath32 , and @xmath33 , where @xmath34 . here
@xmath35 and @xmath36 are the usual critical exponents which , for classical fluids , are those of the three - dimensional ising model , and @xmath37 and @xmath38 are the corresponding nonuniversal amplitudes of the correlation length along the @xmath39 and @xmath2 axes .
when @xmath29 becomes comparable to the characteristic dimension of the system , say the separation @xmath0 between the objects , the size dependence of the thermodynamic functions enters into the thermodynamic potentials through the ratio @xmath40 , and takes a scaling form given by the finite - size scaling theory @xcite .
then _ i ) _ for the interaction of a colloidal particle with spherical shape with radius @xmath4 with a half - space bounded by a plane in the case of a system governed only by short - ranged interactions the theory @xcite predicts : @xmath41 where @xmath42 are the temperature and field relevant in renormalization group sense scaling variables and @xmath43 .
note that this prediction shall be valid for any dimensionality @xmath28 since it simply takes into account that the dimension of a hypersphere in a @xmath28 dimensional space is one dimension less than that of the space itself .
when the hyperscaling holds , i.e. , for @xmath44 for short ranged systems , the scaling function @xmath45 is universal .
when @xmath46 a nonuniversal system dependent prefactor is expected to be present which shall be taken into account in the corresponding analysis .
_ ii ) _ for the interaction mediated by the critical fluid @xcite in the case of two colloidal particles with radii @xmath5 and @xmath6 one has @xmath47 with @xmath48 . in the remainder we are going to study @xmath49 and @xmath50 in a system governed not by
short - ranged , but by long - ranged dispersion interactions .
that will require some modifications of eq .
and eq . .
furthermore , we will obtain @xmath49 and @xmath50 utilizing the derjaguin @xcite and the recently introduced surface integration approach @xcite .
for that aim we will need the corresponding results for a system with a film geometry that is governed by the same dispersion interactions as those of sphere - plate and sphere - sphere interactions .
that is why we are next going to concisely recall these topics .
we start with the behavior of the thermodynamic casimir force in nonpolar fluid systems with dispersion forces .
the specifics of the scaling theory for systems with dispersion forces are described in details in refs .
that is why we will just briefly remind here some very basic facts that will also serve us for introducing notations needed further in the main text .
we consider a fluid systems consisting of a nonpolar medium @xmath23 comprising two thick parallel plates of materials @xmath24 and @xmath25 which are coated by thin solid films of substances @xmath26 and @xmath27 , respectively see fig .
[ fig : fluid_systems]*a*. let @xmath0 denote the distance between the plates , and let us suppose that the thicknesses of the coating films is negligible . if the fluid medium is in contact with a particle reservoir with a chemical potential @xmath2 , the grand canonical potential @xmath51 per unite area @xmath52 of this medium in excess to its bulk value @xmath53 depends on @xmath0 and , thus , one can define the total effective force @xmath54 , which includes , among others , the contributions due to fluctuations of the medium and dispersion interactions in it as @xmath55 in a full accord with eq . , where @xmath56 is the density of the bulk grand canonical potential , and @xmath52 is the surface area of the plates .
we suppose that the dispersion forces governing all the parts of the considered system depend on the distance @xmath57 between the constituents of the bodies in the system @xmath58 , where @xmath28 is the dimensionality of the space and @xmath59 is a parameter that controls the decay of the dispersion interactions . the last implies @xmath60 interactions between the fluid particles and substrate potentials @xmath61 acting on the fluid particles at a distance @xmath62 from the surface of the colloid particles . when @xmath63 such long - range interactions are termed subleading long - ranged interactions @xcite .
the systems governed by them do also belong to the ising universality class characterized by short - ranged forces @xcite , i.e. , the critical exponents , e.g. , do not depend on @xmath59 for such type of interactions . for @xmath64 and
@xmath65 one has the usual van der waals interactions , while @xmath64 and @xmath66 corresponds to the retarded casimir - polder one .
these two interactions are two prominent representative of the class of subleading long - ranged interactions . when @xmath64 , the ising universality class is characterized by critical exponents @xcite @xmath67 clearly , by varying the ratio between the strengths of the long - ranged @xmath68 and the short - ranged
@xmath69 contributions in the fluid interaction one can quantitatively probe the importance of the long - ranged parts of the interactions within the fluid medium .
one can also in this way study potential experiments in colloidal systems which allow for a dedicated tailoring of the form of the effective interactions between colloidal particles .
the contribution of the dispersion forces to the total effective force eq . can be distinguished from that of the ccf by their temperature dependence , because the leading such of the former does not exhibit a singularity .
thus , one can perform the decompositions and identifications given in eq . , eq .
and eq . .
if the system is away from its bulk critical point for the occurring force @xmath70 per cross section area @xmath52 _ _ @xmath71 is customary to write the following expression @xmath72 where , for dimensional reasons , the microscopic length scale @xmath29 is introduced .
let us note that one normally considers the case @xmath73 and , thus , omits the apparent dependence of this length , that can be taken to be , e.g. , the so - called _ retardation length _
@xcite @xmath74 . in eq .
@xmath75 is the hamaker term , whose dependence from the temperature and chemical potential is given by the so - called hamaker constant @xcite ( for details see the appendix in ref .
[ ] ) . as already explained above , upon approaching the bulk critical point of the system the fluctuations of the order parameter of the confined fluid medium exhibit strong correlations which gives rise to new contribution to the ntf , the ccf ( see below ) . in the vicinity of this point ( critical region ) eq . is no longer valid . following ref .
. then becomes @xmath76 \nonumber\\ & & + ( \sigma-1)\beta h_a(t,\mu ) l^{-\sigma}\xi_{\mathrm{ret}}^{\sigma - d}.\end{aligned}\ ] ] here @xmath77 is dimensionless , universal scaling function , @xmath78 and @xmath79 are the temperature and field relevant scaling variables , respectively , while @xmath80 , @xmath81 and @xmath82 are the irrelevant in renormalization group sense scaling variables associated with the interactions in the system . as reported in sec .
ii of ref .
[ ] , for systems belonging to three - dimensional ising universality class with `` genuine '' ( nonretarded ) van der waals interaction @xmath83 , for the correction to scaling exponents one has @xmath84 , @xmath85 , and @xmath86 . in eq .
@xmath87 is the ( dimensionless ) scaling field associated with the wegner - type corrections , while @xmath88 and @xmath89 are dimensionless nonuniversal coupling constants : @xmath88 is proportional to the strength @xmath90 of the long - range part of the interaction potential between the particles of the fluid , whereas @xmath89 are proportional to the contrast between the potentials of the colloids and those in the fluid medium ( see below).we recall that within the mean - field theory with @xmath91 and @xmath92 one has , instead , @xmath93 and @xmath94 .
this implies @xmath95 where all critical exponents take their mean - field values @xmath96 , @xmath97 , @xmath98 . according to the scaling hypothesis of
the ccf one expects that near the bulk critical point @xmath99 where @xmath100 is a scaling function that for large enough @xmath0 with fixed @xmath101 and @xmath102 approaches the scaling function of the short - ranged system @xmath103 ( for details see eqs .
( 2.12 ) and ( 4.10 ) in ref .
[ ] ) . from eqs .
( [ eq : decomp ] ) , ( [ eq : def_casimir ] ) and ( [ eq : def_vdw ] ) , together with eq .
it follows that the scaling function of the ccf @xmath104 is proportional to the sum of @xmath77 and the singular part of the hamaker term @xmath105 .
the last implies that in order to determine the cf in systems with dispersion interaction one has to decompose the contribution captured through the hamaker term in a singular and a regular parts , i.e. @xmath106 thus , with @xmath73 one has @xmath107,\ ] ] while eq .
, with the normalization per unit area , in the general case becomes @xmath108 we will often compare the behavior of the system with subleading long - ranged dispersion interactions present with this one of a system with purely short - ranged interactions which will serve as a reference system .
in such a purely short - range system one has @xmath109 . then
, at the bulk critical point @xmath110 the leading term of the ccf @xmath111 per unit area between the slabs bounding the fluid has the form @xmath112 where @xmath113 . here
@xmath114 is an universal dimensionless quantity , called casimir amplitude , which depends on the bulk and surface universality classes ( and the boundary conditions ) .
since the ccf is proportional to @xmath115 the interaction between the plates can become rather strong in a system with high critical temperature such as , e.g. , in classical binary liquid mixtures .
note that the _ sign _ of the force depends on the sign of the casimir amplitude @xmath116 which , on its turn , depends on the boundary conditions imposed by the bounding surfaces on the fluid . according to the usual convention negative sign corresponds to attraction , while positive sign means repulsion of the surfaces bounding the system .
in what follows , we are going to present results for the ccf , vdwf and ntf in the cases of sphere - sphere and sphere - plate systems , utilizing the knowledge gained from studies of the corresponding interactions between parallel palates . when it comes to calculating some geometry dependent interaction energy or force in systems where at least one of the objects has a nonplanar geometry , the most common approach used is the one first proposed by b. derjaguin @xcite .
it is know as derjaguin approximation in colloidal science ( see e.g. ref . [ ] and p. 34 in ref . [ ] ) , and proximity force approximation in studies of qed casimir effect ( see e.g. p. 97 in ref . [ ] ) .
the main idea behind the da is that one can relate the knowledge for the interaction force / potential between two parallel plates with the one between two gently curved colloidal particles , when the separation between them is much smaller than the geometrical characteristics of the particles in question .
more specifically , the da states that in @xmath64 the interaction force @xmath117 between two spherical particles with radii @xmath118 and @xmath119 placed at a distance @xmath120 is given by @xmath121 where @xmath122 is an effective radius and @xmath123 force per unit area between parallel plates . when the sphere with radius @xmath124 interacts with a plate one has @xmath125 and then eq .
is still valid with @xmath126 .
now if one takes the integrand in eq . in the form given by eq
. with @xmath64 ,
i.e. @xmath127 and performs the integration there , the result is @xmath128 here we recall that the hamaker term @xmath75 , depends both on the dimensionality @xmath28 of the system and the decay parameter @xmath59 characterizing the strength of the van der waals interactions [ see eqs .
( [ ha_ns ] ) and ( [ ha_sing ] ) below ] . as far as the ccf is concerned , the da for @xmath64 , after substituting the force per unit area in the form eq . , the integration of eq
. delivers @xmath129{\rm d}z\nonumber\\ & & \equiv x_{\mathrm{cas , da}}(x_t , x_\mu,\cdots),\end{aligned}\ ] ] where , due to the rapid decay of the interaction , the upper limit of integration has been set to infinity .
of course , this can be justified only if @xmath0 is much smaller than the characteristic sizes of the interacting objects involved .
+ an improvement and generalization of the da , called `` surface integration approach '' has been proposed in ref .
it has been used there to study van der waals interactions between objects of arbitrary shape and a plate of arbitrary thickness .
it delivers _ exact _ results if the interactions involved can be described by pair potentials .
the main advantage of this approach over the da is that one is no longer bound by the restriction that the interacting objects must be much closer to each other than their characteristic sizes .
the main result is that for the force acting between a @xmath130 object ( say a colloid particle ) @xmath131 of general shape @xmath132 and a flat surface bounded by the @xmath133plane of a cartesian coordinate system , one has @xmath134
\mathrm{d}x \mathrm{d}y\nonumber \\ & -&\int_{a_{s}^{\rm away}}\int f^{\parallel}_{{\cal a}}[{\rm s}(x , y ) ] \mathrm{d}x \mathrm{d}y,\end{aligned}\ ] ] where @xmath135 is the projection of the surface @xmath136 of the particle over the @xmath133 plane , @xmath137 .
. has a very simple intuitive meaning : in order to determine the force acting on the particle one has to subtract from the contributions stemming from surface regions @xmath138 that `` face towards '' the projection plane those from regions @xmath139 that `` face away '' from it , where @xmath138 and @xmath139 are the projections of the corresponding parts of the surface of the body on the @xmath133plane .
it is clear that if one takes into account only the contributions over @xmath138 one obtains expression very similar to the da .
the both expressions in that case will differ only by the fact that while eq
. takes into account that the force on a given point of the @xmath136 is _ along the normal _ to the surface at that point , the standard da does not take this into account .
let us recall that eq
. provides exact results for the interaction under the assumption that the constituents of the body interact via pair potentials .
this is , of course , _ not _ the case of ccf .
it is , however , clear that under mechanical equilibrium of the colloid in the fluid , the ccf is again along the normal to the surface at the point of the surface where it acts .
thus , one can get a reasonably good approximation to the effect of that force by keeping just the integration on that part of the surface of the body that faces the plane .
this leads to @xmath140 \mathrm{d}x \mathrm{d}y.\end{aligned}\ ] ] in the simplest case of interaction between a spherical particle and a thick plate [ see fig .
[ fig : siafig]@xmath141 , induced by point - like sources , eq
. takes the form @xmath142f_{{\cal a}}^{\parallel}(z)\mathrm{d}z.\ ] ] the term @xmath143 $ ] in the integrand of eq .
reflects how the projection to the normal to the surface of the sphere changes as a function of @xmath62 . substituting eq . in the above expression for the van der waals sphere - plate interaction
we can write @xmath144.\end{aligned}\ ] ] the corresponding expression for the ccf arising between a sphere and a plate , following eq .
, can then be written as @xmath145f_{\rm cas}^{\parallel}(z){\rm d}z\\ & & = 2\pi r\int_{l}^{l+r}\dfrac{1}{z^{3}}\left[1-\dfrac{z - l}{r}\right]x_{\mathrm{cas}}^{\parallel}[x_t(z),x_\mu(z),\cdots]{\rm d}z.\nonumber\ \ \ \ \\end{aligned}\ ] ] when one is interested in the interaction between two or more objects with nonplanar geometry a general expression like eq .
, that determines the interaction between these objects in terms of the interaction between parallel plates , i.e. , some sort of a sia approach when neither of the bodies has a planar geometry , is currently , as far as we are aware of , not known .
an attempt in this direction was reported in ref .
[ ] , but the equation offered there leads to energy of interaction between two spheres which differs from the classical one reported by hamaker ( see fig . 3 in ref .
[ ] ) . in the current article we will show that in this special case of pair of spherical particles [ see fig .
[ fig : siafig]@xmath146 with radii @xmath118 and @xmath119 one can write the force of interaction between two spherical colloid particles in the form ( for details see appendix [ sec : appspspsia ] ) @xmath147\zeta(z_{2})\mathrm{d}z_{2}\nonumber\\ & & + \int_{l}^{l+2r_{2}}\zeta(z_{2})\int_{z_{2}}^{z_{2}+2r_{1}}f_{{\cal a}}^{\parallel}(z_{1})\mathrm{d}z_{1}\mathrm{d}z_{2}.\end{aligned}\ ] ] where the function @xmath148 is defined as @xmath149}{(l+r_{1}+r_{2})}\right\}.\ ] ] from eq . with @xmath65 and @xmath66 one obtains * when @xmath65 @xmath150 this result can be easily verified by simple differentiation with respect to the separation distance , of the potential obtained in ref .
* when @xmath66 @xmath151\right\ } , \end{aligned}\ ] ] where the two polynomials which enter the above expression are + [ polynomialssigma4 ] @xmath152 and @xmath153 + for the ccf between spherical objects after the substitution of eq . in eq . and
properly taking the limits of integration leads to @xmath154\mathrm{d}z_{2}+\int_{l}^{{l+r_2}}\zeta(z_{2})\int_{z_{2}}^{z_{2}+r_{1}}\dfrac{1}{z_{1}^{3 } } x_{\mathrm{cas}}^{\parallel}[x_t(z_{1}),x_\mu(z_{1}),\cdots]\mathrm{d}z_{1}\mathrm{d}z_{2}.\ \ \ \ \ \ \ \ \ \\end{aligned}\ ] ] since both da and sia utilize the knowledge of the behavior of the force per unit area arising between a pair of parallel palates , in the following section we present the corresponding model within which we describe such a system . because the corresponding expressions have been presented in details in ref .
[ ] , here only key results , as well some notations , are going to be given which will be needed in the remainder of the article .
as explained in ref . [ ] , we consider a lattice - gas model of a fluid confined between two planar plates , separated at a distance @xmath0 from each other , with grand canonical potential @xmath155 $ ] given by @xmath156= k_{b } t \sum_{{\bf r}\in \mathbb{m}}\left\ { \rho({\bf r } ) \ln\left[\rho({\bf r})\right]\right.\nonumber\\ & & \left . + \left[1-\rho({\bf r})\right ] \ln\left[1-\rho({\bf r})\right]\right\ } + \dfrac{1}{2}\sum_{{\bf r } , { \bf r}'\in \mathbb{m}}\rho({\bf r})w^{l}({\bf r}-{\bf r}')\rho({\bf r } ' ) \nonumber \\ & & + \sum_{{\bf r } \in \mathbb{m}}\left[v^{(s_{1}|l|s_{2})}(z)-\mu\right ] \rho({\bf r}),\end{aligned}\ ] ] where @xmath23 is a simple cubic lattice in the region occupied by the fluid medium @xmath157 $ ] and @xmath158 is an external potential that reflects the interactions between the confining plates and the constituents of the fluid . in eq .
@xmath159 is the nonlocal coupling ( interaction potential ) between the constituents of the confined medium and @xmath2 is the chemical potential .
all length scales here and in the remainder are taken in units of the lattice constant @xmath160 , so that the particle number density @xmath161 becomes a number which varies in the range @xmath162 $ ] .
we recall that in the framework of a mean - field treatment with respect to the critical behavior the effective spatial dimension is @xmath163 , irrespective of the actual dimension of the model under consideration . in eq .
@xmath164 where @xmath165 , with @xmath166 and @xmath167 is the discrete delta function ; @xmath168 is a proper lattice version of @xmath169 as the interaction energy between the fluid particles , where @xmath170 is the heaviside step function with the convention @xmath171 . the functional eq
can be considered as a modification of the model utilized by fisher and nakanishi @xcite in their mean - field investigation of systems governed by short - range forces .
taking into account the translational symmetry of the system along the bounding surfaces , the variation with respect to @xmath161 leads to an equation of state for the equilibrium density @xmath172/2 $ ] , where @xmath173 , with @xmath174 is the local order parameter profile @xmath175 . in terms of @xmath176 the equation of state
can be written in the following form @xmath177=\dfrac{\beta}{2}\left[\delta\mu-\delta v(z)\right]\nonumber\\ & & + k\left\{\text{\ { a}}_{d,\sigma}\left(\lambda\right)\phi^{*}(z)+\text{\ { a}}_{d,\sigma}^{nn}\left(\lambda\right)\left[\phi^{*}(z+1)+\phi^{*}(z-1)\right]\right.\nonumber\\&&\left.+\lambda \sum_{z'=0}^{l}g_{d,\sigma}(|z - z'|)\theta(|z - z'|-1)\phi^{*}(z')\right\},\end{aligned}\ ] ] where @xmath178 , @xmath179 , @xmath180 and @xmath181 with @xmath182 .
the functions @xmath183 , @xmath184 and @xmath185 are determined in eqs .
( c10 ) , ( c11 ) and ( c12 ) of ref .
[ ] , respectively . the excess grand canonical potential @xmath186 per unit area , where @xmath52 @xmath187-l\omega_{\rm bulk}$ ] , has the form @xmath188-\dfrac{1}{2}\ln\left[1-\phi_{b}^{2}\right]\right.\nonumber\\&&\left .
+ \dfrac{1}{4}\phi^{*}(z)\ln\left[\dfrac{1+\phi^{*}(z)}{1-\phi^{*}(z)}\right]-\dfrac{1}{4}\phi_{b}\ln\left[\dfrac{1+\phi_{b}}{1-\phi_{b}}\right]\right .
\nonumber\\&&\left.+\dfrac{\beta}{2}\delta v(z)\phi^{*}(z)-\dfrac{\beta\delta\mu}{2}[\phi^{*}(z)-\phi_{b}]\right\}+\beta\omega_{\rm reg},\ \ \ \ \\end{aligned}\ ] ] where @xmath189\nonumber\\ & & \times l^{-\sigma+1}\xi_{\rm ret}^{\sigma - d}.\end{aligned}\ ] ] here @xmath190 is the bulk value of the order parameter ( see eq . ( 4.2 ) in ref .
[ ] ) @xmath191 and @xmath192 are the values of the plates - fluid coupling parameters evaluated at the bulk critical point of the system @xmath193^{-1 } , \mu=\mu_{c}=-2\sum_{\bf r}j^{l}({\bf r})\}$ ] , where the sum is running over the whole lattice . the _ effective _ surface potential @xmath194 in eqs .
( [ order_parameter_equation_d4 ] ) and ( [ deltaomegaexz_equilibrium ] ) is given by @xmath195 where @xmath196 and @xmath197 are the ( @xmath1- and @xmath2-independent ) dimensionless plates - fluid coupling parameters .
contributions of the order of @xmath198 , @xmath199 , etc . , have been neglected .
eq . demonstrates that @xmath89 result from the difference between the relative strength of the substrate - fluid interactions for substrates with density @xmath200 and that of the fluid - fluid interactions for a fluid with density @xmath201 . in eq . the restriction @xmath202 holds because we consider the layers closest to the substrate to be completely occupied by the liquid phase of the fluid , i.e. , @xmath203 , thus ensuring the @xmath3 boundary conditions .
physically this can be accomplished by choosing a proper coating of the surfaces of the plates .
mathematically , this is achieved by taking the limits @xmath204 .
the coupling parameter @xmath205 probes the importance of the long - ranged parts of the interaction potential within the fluid medium @xmath206 in eq .
@xmath207 , i.e. , @xmath208 corresponds to plates `` preferring '' the liquid phase of the fluid , while @xmath209 , or @xmath210 mirrors the one with affinity to its gas phase . if the interactions between the constituents of the fluid are of lennard - jones type one has @xmath211 , as commented in refs .
the marginal case @xmath212 together with @xmath213 describes a pure short - range system ( for details refer to eqs .
( 4.10 ) and the text therein in ref . [ ] ) .
provide the basis of the properties of the model that will be used to determine the sphere - plane and sphere - sphere interactions in the remainder .
using eqs .
( [ deltaomegaexz_equilibrium ] ) and ( [ def ] ) for the net _ total _ force @xmath70 ( per unit area @xmath52 and @xmath71 ) acting between parallel plates bounding the fluid medium the following expression can be written @xmath214\nonumber\\ & & -\dfrac{4ks_{1,c}s_{2,c}}{g(d,\sigma)k_{c}^{2}\lambda}l^{-\sigma}\xi_{\rm ret}^{\sigma - d},\end{aligned}\ ] ] where the last term represents the direct interaction between the plates ( for details see the appendix in ref .
[ ] ) . on the other hand , if one subtracts from the potential @xmath186 its regular part @xmath215 , i.e. , if we consider the quantity @xmath216-l\omega_{\rm bulk},\ ] ] then , in accord with eqs .
( [ eq : decomp ] ) ( [ eq : def_casimir ] ) , the @xmath0 dependence of @xmath217 via eq . provides the singular part of the total force , i.e. , @xmath218 .
explicitly , one has @xmath219.\end{aligned}\ ] ] thus , near @xmath8 the ntf and the ccf are related via the expression @xmath220 where the last term is the mathematical equivalent of eq . per unit area @xmath52 and @xmath71 for a film geometry with @xmath221 being the nonsingular ( regular ) part of the hamaker term given by @xmath222.\ \
\ \ \ \ \ \ \ \\end{aligned}\ ] ] for the singular part of @xmath75 within the presented model the following expression was derived [ see eq .
( a6 ) in ref .
[ ] ] @xmath223 where the @xmath1 and @xmath2 dependance is carried by the bulk order parameter @xmath190 [ see eq .
( 4.2 ) in ref . [ ] ] . in the next section ,
based on the results reported in sections [ sec : thermalcas ] and [ sec : model ] , we present numerical results for the behaviour of the above discussed forces in sphere - plate and sphere - sphere fluid systems for the cases @xmath83 . in subsec .
[ subsec : experimentalfeas ] we also show that the values of the parameters used in the numerical evaluations can be experimentally achieved for specific properly used materials .
in the current section , using the results for @xmath91 from the mean - field type numerical study discussed in sections [ sec : thermalcas ] and [ sec : model ] , we will present some approximate results for the behavior of the ccf , vdwf and ntf between two spherical particles as well as between a spherical particle and a plate in @xmath64 , using both the da and sia approximations .
as it became clear from the above shown equations , the key knowledge which is required for the desired calculations to be performed , is the force [ scaling function(s ) ] per unit area between two parallel semi - infinite spaces ( plates ) , for many different separations @xmath0 , and at various values of the temperature @xmath1 and the chemical potential @xmath2 of the fluctuating fluid medium . within the mean - field theory the dependance of the corresponding forces from the temperature @xmath1 and the chemical potential difference @xmath224 near the bulk critical point @xmath225
is given by the dimensionless temperature and field scaling variables @xmath226 and @xmath227 . in our numerical treatment
we take these variables to range in the intervals : @xmath228\equiv[-576;576]$ ] and @xmath229\equiv[-13824;13824]$ ] . for the study of the scaling function of the ccf within the da , the separation distance @xmath0 between the set of parallel plates
is varied from 20 to 100 with step 10 and from 100 to 1000 with step 100 .
in addition , within the sia approximation , the gap length @xmath0 was varied from 20 to 60 with a step of 2 , from 60 to 100 with a step of 5 , and from 100 to 200 with a step of 10 . in order to demonstrate the general tendencies in the behaviour of the scaling functions , we consider one of the plates - fluid coupling parameters fixed ( say @xmath230 ) , having a value either @xmath231 or @xmath232 , while the other one , @xmath233 , is varied from @xmath231 to @xmath232 with step @xmath234 .
the fluid - fluid coupling parameter @xmath205 is supposed to be either @xmath231 or @xmath235 .
+ + from eqs .
( [ daspsp ] ) and ( [ intdavdw ] ) considering only genuine van der waals interactions @xmath236 in @xmath64 one has that within the da the forces are calculated via the following expression @xmath237 where in the case of sphere - plate interaction @xmath238 @xmath43 and @xmath239 , with @xmath240 when the interaction is between spheres @xmath241 .
analogically for the ccf using eqs .
( [ daspsp ] ) and ( [ intdacas ] ) one can write @xmath242 where the scaling function @xmath243 is calculated as follows @xmath244{\rm d}\bar{z}\nonumber\\ & & + \sum_{i = j+1}^{j_{\max}-1}\int_{\lambda_{i-1}}^{\lambda_{i}}\bar{z}^{-3}x_{\rm cas}^{\parallel}[\mathfrak{a}_{i}(\bar{z})]{\rm d}\bar{z}\nonumber\\ & & + \int_{\lambda_{\max}}^{\infty}\bar{z}^{-3}x_{\rm cas}^{\parallel}[\mathfrak{a}_{\max}(\bar{z})]{\rm d}\bar{z},\end{aligned}\ ] ] where @xmath245 ; @xmath246 is dimensionless variable ; @xmath247 is the largest system for which numerical data are available and the arguments of the scaling function are given by @xmath248\triangleq[x_{t}(l_{z;j}),x_{\mu}(l_{z;j}), ...
]$ ] , with @xmath249 . in the calculations performed in the current article @xmath250 . here
@xmath104 are normalized according to eq .
( 5.1 ) from ref .
[ ] so that they contribute correctly to the ccf and ntf in @xmath251 .
as stated at the beginning of the article , the boundary conditions of interest are those of strong adsorption , i.e. @xmath3 boundary conditions . within the sia the corresponding genuine vdwf in a sphere - plate and sphere - sphere systems
is estimated with the use of eqs .
( [ spplgeneralsia ] ) with @xmath65 @xmath252 and eq . ,
respectively . for the ccf in the sphere - plate case ,
using eq . , one has @xmath253 where the scaling function @xmath254 for the sphere - plate ccf is calculated as follows @xmath255{\rm d}\bar{z}+\sum_{i = j+1}^{j_{\max}-1}\int_{\lambda_{i-1}}^{\lambda_{i}}i^{r,|}(\bar{z}|\lambda_{i})x_{\rm cas}^{\parallel}[\mathfrak{a}_{i}(\bar{z})]{\rm d}\bar{z}+\int_{\lambda_{\max}}^{1+\xi}i^{r,|}(\bar{z}|\lambda_{\max})x_{\rm cas}^{\parallel}[\mathfrak{a}_{\max}(\bar{z})]{\rm d}\bar{z},\end{aligned}\ ] ] where @xmath256\theta(\xi-\lambda_{j}),\ ] ] with @xmath257 being the heaviside step function with the convention @xmath171 . in the case of sphere - sphere geometry the expressions for calculating ccf between them within the sia , using eq .
, is @xmath258 where @xmath259{\rm d}\bar{z}_{2 } + \sum_{i = j+1}^{j_{\max}-1}\int_{\lambda_{i-1}}^{\lambda_{i}}i_{1}^{r_{1},r_{2}}(\bar{z}_{2}|\lambda_{i})x_{\rm cas}^{\parallel}[\mathfrak{a}_{i}(\bar{z}_{2})]{\rm d}\bar{z}_{2}\nonumber\\ & & + \int_{\lambda_{\max}}^{1+\xi_{2}}i_{1}^{r_{1},r_{2}}(\bar{z}_{2}|\lambda_{\max})x_{\rm cas}^{\parallel}[\mathfrak{a}_{\max}(\bar{z}_{2})]{\rm d}\bar{z}_{2}+\int_{1}^{1+\xi_{2}}\left\{\int_{1}^{\lambda_{j}/\bar{z}_{2}}i_{2}^{r_{1},r_{2}}(\tilde{z},\bar{z}_{2}|\lambda_{j})x_{\rm cas}^{\parallel}[\mathfrak{a}_{j}(\tilde{z})]{\rm d}\tilde{z}\right .
\nonumber\\ & & \left.+\sum_{i = j+1}^{j_{\max}-1}\int_{\lambda_{i-1}/\bar{z}_{2}}^{\lambda_{i}/\bar{z}_{2 } } i_{2}^{r_{1},r_{2}}(\tilde{z},\bar{z}_{2}|\lambda_{i } ) x_{\rm cas}^{\parallel}[\mathfrak{a}_{i}(\tilde{z})]{\rm d}\tilde{z}\right.\left.+\int_{\lambda_{\max}/\bar{z}_{2}}^{1+(\xi_{1}/\bar{z}_{2})}i_{2}^{r_{1},r_{2}}(\tilde{z},\bar{z}_{2}|\lambda_{\max})x_{\rm cas}^{\parallel}[\mathfrak{a}_{\max}(\tilde{z})]{\rm d}\tilde{z}\right\}{\rm d}\bar{z}_{2},\ \ \ \ \ \ \ \\end{aligned}\ ] ] with @xmath260 , [ ir1r2 ] @xmath261 @xmath262 @xmath263}{(1+\xi_{1}+\xi_{2})^{2}},\ \ \ \ \ \ \ \ \end{aligned}\ ] ] and the arguments of the scaling function are defined via @xmath264\triangleq[x_{t}(l_{z_{k};j}),x_{\mu}(l_{z_{k};j}), ... ],\ k=1,2;\ j=1,j_{\max}$ ] .
after presenting the mathematical means to calculate the ccf and vdwf , we now pass to the detailed discussion of the results and argumentation of the experimental feasibility of the parameters used in the model calculations utilizing da and sia .
we start with the simplest case of systems ( any of the depicted in fig .
[ fig : fluid_systems ] ) governed by purely short - range interactions , i.e. @xmath265 . if @xmath266 is the scaling function of the casimir force in this case it is clear the ccf is simply @xmath267 .
it results only from the correlated fluctuations and the size dependence of the spatial order parameter @xmath268 .
following eqs .
( [ casimirf_l ] ) , ( [ ha_ns ] ) and ( [ ha_sing ] ) we see that @xmath269 coincides with @xmath270 . irrespective of the geometry of the interacting objects @xmath269 is _ negative _ , which corresponds to _ attractive _ force , at any separation @xmath0 and for any value of the scaling variables @xmath271 and @xmath272 under @xmath3 boundary conditions . for sphere - plate and sphere - sphere systems
the study of @xmath273 within the da for different separations between the interacting objects showed that when @xmath274 the curves fall on that obtained via eq .
, see figs .
[ fig : xcrits11s2gz ] @xmath275 and @xmath276 and fig.[fig : xcrits11s2sz ] @xmath275 and @xmath276 , using : _
i ) _ the exact analytical results based on eqs .
( 4.10 ) reported in ref .
[ ] when @xmath277 ; _ ii ) _ the data for the plate - plate interaction calculated within the presented mean - field theory for @xmath278 .
note that the scaling function of the ccf ( and @xmath273 respectively ) , see eq .
, as well as the vdwf , see eq .
, is _ one and the same _ for sphere - plate and sphere - sphere systems , irrespective of the interaction type which takes place between the constituents of a considered system , when the behaviour is studied within the da [ see eqs .
( [ davdwspspsppl3d ] ) ( [ scalfunda ] ) ] . as it is clear from eqs .
( [ davdwspspsppl3d ] ) ( [ scalfunda ] ) , the corresponding forces differ _ only _ by a multiplication factor which depends on some basic geometrical characteristics of the interacting objects .
staying within the da , we study the behaviour of @xmath279 for a system with long - range interactions characterized by @xmath280 , @xmath281 and @xmath282 .
we see that practically all curves except that for @xmath283 coincide with @xmath284 as depicted in fig .
[ fig : xcrits11s2gz]@xmath275 and @xmath276 . upon increasing @xmath233
the curves disperse and all appear below @xmath284 . for @xmath285
all scaling functions are clearly distinguishable from one another [ see fig .
[ fig : xcrits11s2gz]@xmath286 and @xmath287 .
this is easy to understand , given that the nonnegative values of the coupling parameters @xmath288 and @xmath205 are associated with an additional enhancement of the ordering in the system , both near the surfaces of the interacting bodies and in the bulk of the system .
considering eq .
( 2.12 ) in ref .
[ ] we see that at small distances the nonuniversal behaviour dominates and adds to the universal one ( described by @xmath270 ) with a positive sign , resulting in net scaling functions with minima deeper then that of a short - range system . as expected , when the distance is increased the influence of the long - range interactions decreases in the bulk of the system , and as a result all scaling functions tend to the _
universal one_. in this line of thinking for systems characterized by @xmath289 and @xmath290 we expect @xmath291 to remain again negative for any separation @xmath0 and at any value of the scaling variables , irrespective of the value of @xmath205 ( which is always nonnegative ) .
indeed , this turns out to be true and is depicted in fig . [
fig : xcrits11s2sz](*a * ) and ( * b * ) .
however when @xmath292 and the separation between the walls is relatively small a significant part of the system is disordered which results in nonnegative or sign - changing scaling function [ see fig .
[ fig : xcrits11s2sz](*c * ) and ( * d * ) ] . as the distance @xmath0
is increased the influence of the effective surface potential @xmath293 [ see eq .
] quickly decreases and only the additional ordering effect of the fluid - fluid interactions influences the behavior of the order parameter and hence of @xmath291 .
when both wall - fluid coupling parameters are negative , @xmath291 is negative for any @xmath294 and @xmath295 only for very large separations @xmath0 where the effect of the long - ranged interactions on the behavior of the system is negligible . naturally , since the short - ranged surface potentials do support @xmath3 boundary conditions , the role of the negative substrate potentials , which oppose the order near the boundary , will be stronger than that of the positive substrate potentials which try to reinforce the phase preferred near the boundary . for example , we observe that the behavior of @xmath291 in a system with @xmath296 and in such with @xmath297 is almost identical for any @xmath0 as a function of @xmath295 [ compare fig .
[ fig : xcrits11s2sz](*d * ) and fig .
[ fig : xcrits1m1s2sz](*b * ) ] .
thus , for a fixed @xmath205 the behavior of the scaling function is mainly determined by the interplay between the short - ranged surface fields and the strong negative wall - fluid coupling @xmath233 .
if @xmath298 and @xmath299 the scaling function @xmath291 exhibits an unexpected behavior as a function of @xmath0 : for moderate values of @xmath0 in the range @xmath300 to @xmath301 , the maximum of the _ repulsive _ part of the force increases with increasing @xmath0 both as a function of @xmath294 and @xmath295 [ see fig .
[ fig : xcrits1m1s2sz](*c * ) and ( * d * ) ] ; for larger values of @xmath0 the maximum decreases , as expected , and the overall behavior of the scaling function approaches that one of the system with completely short - ranged interactions . so far the discussion was focused on the behavior of @xmath273 within the da , i.e. for separations @xmath0 much smaller than the characteristic geometrical extend of the interacting objects [ in the context of the current article , the sphere(s ) radius(ii ) ] . the quantitative and qualitative comparison in the behavior of the scaling functions within the da and sia is presented on fig . [ fig : xcritsia ] .
let us emphasize again that in contrast to the da , the use of sia _ does not _ put any restrictions on the sizes and separation lengths that can be considered .
in order to illustrate only the main idea , here we restrain ourselves to the choice of parameters @xmath302 , @xmath303 and @xmath282 in a system with @xmath283 fluid layers .
the first distinction one notices is that in comparison to the da , within the sia the mathematical expressions for calculating the ccf and vdwf are different for the sphere - palate and sphere - sphere systems , respectively - see the corresponding equations in subsecs .
[ sec : spplandspspgeomforces ] and [ subsec : spplspspdasiad3 ] : when studying @xmath273 , for the interaction between a planar substrate and a spherical particle eq .
is used , while between a pair of spherical particles one uses eq . .
the results presented on fig .
[ fig : xcritsia](*a * ) and ( * b * ) demonstrate that with the increase of the sphere s radius to length separation ratio @xmath304 the overall behaviour of the scaling functions tends to that predicted by the da .
one also notices that while the curve evaluated within the da corresponds to repulsive force in the entire interval of values of @xmath294 and @xmath295 , when the ration @xmath304 is finite @xmath305 change sign once for @xmath306 [ fig .
[ fig : xcritsia](*a * ) ] and @xmath307 , with a shallow attractive minimum , which gradually tends to zero with increase of @xmath304 .
more specifically , studying the temperature dependance of the scaling function we notice that while those evaluated within the da exhibits a single minimum and two maxima of equal height , the maxima of the scaling function evaluated for a system with @xmath308 differ with about 10 % .
another comparison shows that the difference between the global maximum of @xmath305 evaluated for @xmath308 and any of the two of @xmath291 is nearly 2 times . as a function of the field scaling variable [ fig .
[ fig : xcritsia](*b * ) ] the curve corresponding to @xmath308 exhibits double _ sign change _ in the `` gas phase '' of the fluid medium .
for @xmath304 just under @xmath309 the minimum of the scaling function is zero . for @xmath310 all curves obtained for @xmath311
describe repulsive force . also with increase of @xmath304
the value of @xmath295 at which the minimum occurs goes to zero , while that of the global maximum changes slightly . for the field dependance of
@xmath273 one finds that the difference between the global maxima of @xmath312 and @xmath291 is 73 % . for the study of the ccf between two spherical particles within the sia we choose to vary the ratio @xmath313 and fix that of @xmath314 to 1
- see fig .
[ fig : xcritsia](*c * ) and ( * d * ) .
the depicted on those two subfigures justifies the mathematically predicted limit @xmath315 , both as a function of @xmath294 and @xmath295 . on fig .
[ fig : xcritsia](*c * ) one sees that for moderate values of @xmath313 , the scaling function exhibits 3 maxima ( one in each phase and at the critical point ) and a couple of minima . with the increase of @xmath313
the maximum at @xmath316 and the minimum which appears in the `` liquid '' phase both disappear , as the scaling function of a sphere - sphere system with @xmath317 approaches that for the sphere - plate system with @xmath318 . on a quantitative level , as a function of @xmath294 the difference between the values at the global maxima of @xmath319 and @xmath320 is about 3 times .
the scaling functions evaluated for various values of @xmath313 and @xmath295 differ only quantitatively . in analogy with the comparison made for @xmath321 , here we have that the difference between the values at the global maxima as well as minima of @xmath319 and @xmath320 is about 2.4 times .
we will close the discussion with some comments on the behavior of the ntf , which results from the interplay between the ccf and vdwf in a sphere - palate and sphere - sphere systems studied within the sia .
the obtained results are depicted on fig .
[ fig : tfsia ] . in scope of arguing the experimental feasibility of the presented theory , which will be done in the following section , here the comment will be put on a particular system with @xmath283 characterized by the following coupling parameters : @xmath280 , @xmath322 and @xmath323 .
the behavior of the hamaker term ( associated with the vdwf ) both as a function of @xmath294 and @xmath295 is rather trivial , when one considers a lattice - gas model .
an example of it for a system with @xmath324 is depicted on fig . 5 in ref .
unlike the case presented there , when @xmath325 , as considered here , the direct substrate - substrate interaction [ the first term in eq . ] influence the behaviour of the total cf .
for @xmath326 at @xmath327 the hamaker term is non - zero and corresponds to repulsion for any value of @xmath294 . on the other hand , as a function of the field
scaling variable , at @xmath278 and for @xmath326 , the hamaker term changes sign once in the `` gas '' phase of the fluid , being attractive for @xmath328 and repulsive otherwise , with an infinite slope at @xmath329 . with regard to the behaviour of @xmath330 both as a function of @xmath294 and @xmath295 for the system specified above , one observes a single sign change with a low repulsive maximum , appearing in the `` liquid '' phase of the fluid , which gradually tends to zero with increase of @xmath304 .
after superimposing both quantities , namely @xmath331 and @xmath330 , using eqs .
( [ ha_ns ] ) , ( [ ha_sing ] ) , ( [ siavdwsppl3d ] ) , ( [ siacasspplate3d ] ) ( [ ivfun ] ) , one ends up with the temperature [ see fig .
[ fig : tfsia](*a * ) ] and field [ see fig . [
fig : tfsia](*b * ) ] dependencies of the total force between a spherical particle of arbitrary radius and a planar substrate within the sia . in the previous section
it has been argued that if the coupling parameters are tuned in a certain way , one can realise not only repulsive or sign changing ccf , but even the net total force can exhibit such a behaviour .
particular values of the model parameters at which these phenomena are observed were @xmath280 , @xmath322 and @xmath323 . in the current section we will show that there are materials leading to such values of these parameters , thus demonstrating the feasibility of the theoretical predictions presented .
let us consider a sphere - plate and sphere - sphere systems immersed in xenon ( xe ) , which exhibits critical fluctuations near its bulk critical point .
this is obviously a nonpolar simple fluid .
its physical characteristics are presented in table i in ref .
we will make the assumption that all interactions between the constituents of a given system are of lennard - jones type , i.e. , @xmath332 and @xmath333 ( see eq . ( 7.1 ) in ref .
[ ] and the text therein ) . for the calculation of the substrate - fluid coupling constant @xmath334 within the mean - field treatment of the problem we use the following expression : @xmath335 $ ] , where @xmath336 ( see column 3 in table [ table_fluid_sub ] ) and @xmath68 ( see column 7 of table i in ref . [ ] ) are the long - range inter - particle interaction energies and @xmath337 ( see column 7 in table [ table_fluid_sub ] ) is the number density of the bulk substrate relative to the critical one of the fluid medium @xmath201 .
the quantitative assessment of the interatomic interactions between the xenon and some concrete substances listed in column 1 of table [ table_fluid_sub ] , manifests itself with values of @xmath334 ( see column 8 of table [ table_fluid_sub ] ) very near to the above mentioned model ones . with regard to the experimentally realizable spherical particles studied here ,
the following composites are considered : ruthenium @xcite ( ru ) or platinum @xcite ( pt ) core , encapsulated by thin ( about 4 unit cells or @xmath338 ) @xmath339 or @xmath340 film @xcite ; lithium ( li ) core , encapsulated in a carbon shell @xcite ; carbon @xcite ( c ) or silica @xcite ( sio@xmath341 ) aerogels .
the contact surface of any of the interacting geometries ( spherical or planar ) is assumed coated by monolayer of lead @xcite ( pb ) or thallium nitride @xcite ( tln ) to ensure the ( + , + ) boundary conditions . in addition , from the data reported in ref .
[ ] we have that @xmath342 and @xmath343 .
.[table_fluid_sub ] physical characteristics and interaction parameters of the considered core materials : ruthenium ( ru ) , platinum ( pt ) , lithium ( li ) , carbon ( c ) and silica ( sio@xmath341 ) aerogels with the fluid medium ( xe ) . the columns show : the distances @xmath344 and @xmath345 in ( columns 2 and 4 ) at which the inter - particle potential within the substrate and between it and the fluid is zero , the corresponding potential well depths @xmath346 and @xmath347 in units @xmath115 ( columns 3 and 5 ) , the density @xmath348 of the substrates in @xmath349 ( column 6 ) , the number density @xmath337 ( column 7 ) and the substrate - fluid coupling parameter evaluated at the critical temperature
@xmath334 ( column 8) .
the values of @xmath344 and @xmath346 are taken from ref .
[ ] , while those of @xmath345 and @xmath347 are calculated via kong s mixing rules ( see eqs .
( 7.2 ) in ref . [ ] ) .
the densities of the aerogel substrates @xcite are taken in such a way as to render the desired value of @xmath334 , namely @xmath350 . [
cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] [ table ] although the mean - field theory gives poor quantitative estimation of the behavior of the studied forces , it is tempting to evaluate them , nevertheless . as is has been done in the calculations leading to the results reported in fig .
[ fig : tfsia ] , we fix the separation the interacting objects to @xmath351 , where @xmath160 is the suitable distance between the xenon atoms at the bulk critical point ( see column 5 of table i in ref . [ ] ) .
then , studying the temperature dependance of the ntf occurring in a sphere - plate system , see fig . [
fig : tfsia](*a * ) with @xmath352 at @xmath277 , one shell observe a double sign change at @xmath353 and @xmath354 , with an attractive global minimum at @xmath355 , which magnitude is @xmath356 .
if now the interaction takes place between a plate and a sphere with radius @xmath357 , the force sign change will be at @xmath358 and @xmath359 , the minimum will remain attractive , observed at @xmath360 , and with magnitude of @xmath361 . as a function of the chemical potential difference at @xmath278 ,
[ fig : tfsia](*b * ) , the force changes sign only once irrespectively on the sphere s radius , with the occurrence of a single attractive minimum .
thus , when @xmath352 the force is repulsive for @xmath362 and attractive otherwise .
the minimum is reached at @xmath363 , with a magnitude of @xmath364 .
the increase of the sphere s radius 10 times results in insignificant change in the value of @xmath365 at which the sign change occurs , but the magnitude of the force at its minimum increases substantially to @xmath366 , observed at @xmath367 . for even larger spherical particles @xmath368
the value of the force at its minimum is a linear function of the ratio @xmath304 [ see eq . ] and for the particular case considered here can be estimated with the use of the expressions : @xmath369 and @xmath370 .
now we focus our attention on the interaction between spherical particles in critical xenon .
let the radius of one of the particles we consider fixed is , say @xmath371 , and vary the other one @xmath118 .
starting with the temperature dependance fig .
[ fig : tfsia](*c * ) , we see that when @xmath372 the ntf is repulsive at any temperature , only becoming zero at @xmath373 . with the increase of @xmath118 a double sign change and a single attractive minimum
are observed .
for @xmath374 the magnitude of the force s minimum is @xmath375 . at @xmath278 and
@xmath372 one observes not so trivial behaviour of the ntf as a function of the field scaling variable [ see the last paragraph of subsec .
[ subsec : discussion ] as well as fig .
[ fig : tfsia](*d * ) and ( * d * ) ] .
for such a system , the global minimum has a value of @xmath376 and a maximum of @xmath377 .
upon increasing the value of @xmath118 to @xmath378 , the force changes sign only once , having a single attractive minimum with magnitude of @xmath379 . for the sake of completeness , we give the critical temperature of xenon @xmath380 and the value of the critical chemical potential per @xmath115 : @xmath381 , calculated using eqs .
( 2.2b ) , ( 5.37b ) and ( 8.8 ) from ref .
in the current article we have studied the interactions between objects governed by dispersion forces immersed in a nonpolar critical medium also governed by such forces .
we envisage the case in which the critical medium is a nonpolar one - component fluid or a binary liquid mixture , i.e. , its critical bulk behavior belongs to the ising surface universality class . because of the modifications of the order parameter of the fluid , as well as of its fluctuation spectrum , in addition to the dispersion force one also has an additional effective force known as the ccf , acting between these objects .
since the dispersion forces influence the critical medium , they change the order parameter profile and the fluctuations of the system .
thus , studying the ccf in such systems one , unavoidably , studies the interplay between these two long - ranged and fluctuation induced forces .
in addition to the contribution of any of these forces to the overall interaction between the immerse objects we also studied the ntf between them . in order to achieve this
we have used general scaling arguments and mean - field type calculations utilizing the derjaguin and the surface integration approaches see subsec .
[ sec : spplandspspgeomforces ] .
any of these two approaches uses data for the forces between two parallel slabs separated at a distance @xmath0 from each other , made of the same materials as the objects and confining the same fluctuating fluid medium characterized by its temperature @xmath1 and chemical potential @xmath2 . the corresponding model , which we have used to produce the data needed for the current study , in presented in sec . [ sec : model ] .
in the article we concentrated on a system that involves either a sphere and a thick planar slab , or two spheres with , in general , different radii
[ fig : fluid_systems ] .
the surface of any of the objects immersed in the fluid is supposed coated by thin layers exerting strong preference to the liquid phase of the fluid , or one of the components of the mixture , modeled by strong adsorbing local surface potentials ensuring the so - called @xmath3 boundary conditions . we suppose that the core region of the slab and the particles , on the other hand , influence the fluid by long - ranged _ competing _ dispersion potentials . figures [ fig : xcrits11s2gz ] [ fig : xcritsia ]
show that for a suitable set of colloid - fluid , slabs - fluid , and fluid - fluid coupling parameters the competition between the effects due to the coatings and the core regions of the objects involved result in a _
sign change _ of the ccf between the objects considered .
figures [ fig : xcrits11s2gz ] [ fig : xcrits1m1s2sz ] present these results obtained within the da , while fig .
[ fig : xcritsia ] shows the ones obtained within the sia .
as it is clear from the data presented , the sign change happens when one changes the temperature @xmath1 , the chemical potential of the fluid @xmath2 , or the distance @xmath0 between the colloid and the slab or between the colloids .
figure [ fig : tfsia ] shows , on its turn , that a _ sign change _ can also be achieved for the _ net total force _ between these objects .
it is worthwhile to emphasize that this can be used for governing behavior of objects , say colloidal particles , at small distances , say in colloid suspensions for segregation of colloids .
it can also provide a strategy for solving problems with handling , feeding , trapping and fixing of micro parts in nanotechnology resolving the issues related to sticking of the particles on the surface of the mechanical manipulator utilizing , e.g. , the _ reversible dependence _ on the forces under minute changes of the temperature of the critical medium .
one can perform grabbing of particles for small values of @xmath271 , where the force is attractive and release them at a given spacial position after slightly increasing or decreasing of temperature achieving in that way a value of @xmath271 with a repulsive ntf .
data given in table [ table ] for specific substances demonstrate that the values of the parameters used in the theoretical calculations are experimentally feasible .
finally let us make some remarks on the sia which has been suggested in ref .
@xcite as an improvement and generalization of the da .
it delivers an _ exact _ expression between a body with a general shape and a slab , expressing the interaction in terms of the plane - plane interaction , provided the interactions involved can be described by pair potentials , say van der waals type potentials . as eq .
shows , the final result is expressed in terms of integral over projection @xmath138 of the surface of the colloid particle on the plane which faces towards this plane minus the contribution over the surface projection @xmath139 that faces away from the plane .
the main advantage of this approach over the da , which involves only integration over @xmath138 , is that it is not bound by the restriction that the interacting objects must be much closer to each other than their characteristic sizes .
the integral over @xmath138 only can be used , as it is customary under da , to evaluate the interaction between the surface of the colloid and a plane also for the cases when the interaction can not be prescribed to some point like sources .
the last includes , e.g. , the way it is used for calculating the ccf . if the colloid is in a mechanical equilibrium in a fluid it is clear that the force at a given point on the surface of the colloid is orthogonal to the surface at that point
. then one can suggest an improvement over da also for the case of ccf type interactions since the standard da does not respect this fact .
the corresponding improvement is suggested in eq . for a colloid particle of any shape and a plane .
unfortunately , we are not aware for an expression for the interaction of two bodies of general shape expressed in terms of the corresponding plane - plane interactions .
we have been able , however , to derive such an expression for the case of two spheres see eq . .
it has been shown that it delivers correct results for interactions decaying with the distance @xmath58 , with @xmath64 and @xmath382 reproducing , for @xmath65 the classical result of hamaker @xcite , see eq .
, and delivering a new analytical result for the retarded van der waals interaction with @xmath66 see eq . .
furthermore , this expression delivers in the limit @xmath383 the standard derjaguin result eq . for the sphere - sphere interactions for the proof see appandix [ sec : appspspsia ] . the corresponding expression for the ccf between two spheres within the sia approach that takes into account only the interactions between surfaces facing each other , similar to the approach of da ,
is given in eq . .
a comparison of the results obtained via da and sia is presented in figs .
[ fig : xcrits11s2gz ] [ fig : xcritsia ] .
the authors gratefully acknowledge the support of the program for career development of young scientists , bas via contract no .
all numerical data used for the evaluation of the sphere - plate and sphere - sphere interactions were obtained with the use of avitohol
cluster platform of the institute of information and communication technologies at bas .
g.v . expresses its sincere thanks to dr .
k. shterev for the helpful advices on the numerical calculations procedures .
within the _ pairwise _ summation hypothesis @xcite proposed by h. hamaker in 1937 , the van der waals interaction energy @xmath384 between two _ macroscopic _ bodies in @xmath64 can be written as an integral over the individual interactions @xmath385 between all particles constituting the two objects where @xmath0 is the minimal distance between the objects ; @xmath388 are the volumes occupied by the interacting bodies ; @xmath200 are their number densities ; @xmath57 denotes the distance between the elementary volume elements @xmath389 ; @xmath390 is any of the van der waals coupling parameters and @xmath59 is a parameter that characterizes the decay of the interaction ( @xmath65 for the keesom , debye and the nonretarded london - van der waals interactions while @xmath66 for the retarded casimir - polder one ) . therefore within the hamaker approach , for the interaction energy between a sphere of radius @xmath118 and a point - like object @xmath391 at a distance @xmath392 to the sphere s center one can write ( see eq .
( 5 ) in ref . [ ] ) @xmath393{\rm d}r.\ ] ] now by making use of the expression for the interaction energy @xmath394 between a point - like object and a half - infinite space ( plate ) ( for details see eqs .
( 7 ) , ( 9 ) and ( 17 ) in ref .
@xcite ) , eq . can be written equivalently as @xmath395\nonumber\\&&-\dfrac{1}{r } \int_{r - r_{1}}^{r+r_{1}}\omega^{p,|}(z_{1}){\rm d}z_{1},\end{aligned}\ ] ] where @xmath396 using the same considerations as in the derivation of eq . , hamaker managed to obtain for the interaction energy @xmath397 between two spheres of radii @xmath118 and @xmath119 , separated at a distance @xmath398 apart the following expression @xmath399{\rm d}r.\end{aligned}\ ] ] substituting eq . into eq .
, taking into account that @xmath400 and performing differentiation with respect to the separation @xmath0 , one ends up with an expression relating the interaction force @xmath401 between two spherical particles with the force per unit area between two parallel plates where the function @xmath148 is given in eq . in the main text . because in the derivation of eq .
no restrictions on the allowed sizes and separations between the particles are considered , one can say that this expression was derived within the sia .
@xmath403}{z_{2}^{\sigma } } + \dfrac{[(\sigma+1)r_{1}+z_{2}]}{(z_{2}+2r_{1})^{\sigma}}\right\}\zeta(z_{2})\mathrm{d}z_{2}\nonumber\\&&= \dfrac{2h_{a}^{\rm ( reg)}\xi_{\rm ret}^{\sigma-3}}{(\sigma-2)(\sigma-3)(l+r_{1}+r_{2})}\left[\dfrac{{\cal e}(l|\sigma)}{(\sigma-4)(l+r_{1}+r_{2})}+{\cal e}(l|\sigma+1)\right],\end{aligned}\ ] ] from eq .
the general expression describing the @xmath0 dependance of the sphere - plate interaction follows directly .
indeed by taking , say @xmath410 [ i.e. one of the spheres becomes a half - space ( plate ) ] , one has that for any inverse distance force law @xmath411 and @xmath412 .
thus @xmath413 which is exactly the result eq . (
20 ) reported in ref .
it is easy to show that from eq .
follows the expression obtained by derjaguin ( see for example eq . ( 1 ) in ref . [ ] ) describing the interaction between spherical particles at close proximity to one another [ i.e. @xmath414 , or @xmath415 . introducing the notations @xmath416 , eq .
takes the form @xmath417\zeta(\bar{z}_{2}){\rm d}\bar{z}_{2}\nonumber\\ & & + \dfrac{1}{l^{\sigma-2}}\int_{1}^{1 + 2\xi_{2}}\int_{\bar{z}_{2}}^{\bar{z}_{2}+2\xi_{1}}\zeta(\bar{z}_{2})f_{a}^{\parallel}(\bar{z}_{1}){\rm d}\bar{z}_{1}{\rm d}\bar{z}_{2}.\end{aligned}\ ] ] now we notice that within the da @xmath418 then @xmath419 , @xmath420 and @xmath421 .
therefore @xmath422 in a way analogous to the use of the da for determination of the ccf between objects of complicated geometry , one can make use of the sia technique to do the same . then
, as usual , the restriction appears that the integration must be carried on only over these parts of the objects surfaces which are facing each other ( see for details the text above ) .
taking this into account we can write the following expressions for the ccf between two spherical particles of arbitrary radii as well as between a spherical particle and a planar substrate respectively | we study systems in which both long - ranged van der waals and critical casimir interactions are present .
the last arise as an effective force between bodies when immersed in a near - critical medium , say a nonpolar one - component fluid or a binary liquid mixture .
they are due to the fact that the presence of the bodies modifies the order parameter profile of the medium between them as well as the spectrum of its allowed fluctuations .
we study the interplay between these forces , as well as the net total force ( ntf ) between a spherical colloid particle and a thick planar slab , and between two spherical colloid particles .
we do that using general scaling arguments and mean - field type calculations utilizing the derjaguin and the surface integration approaches .
they both are based on data of the forces between two parallel slabs separated at a distance @xmath0 from each other , confining the fluctuating fluid medium characterized by its temperature @xmath1 and chemical potential @xmath2 .
the surfaces of the colloid particles and the slab are coated by thin layers exerting strong preference to the liquid phase of the fluid , or one of the components of the mixture , modeled by strong adsorbing local surface potentials , ensuring the so - called @xmath3 boundary conditions . on the other hand ,
the core region of the slab and the particles , influence the fluid by long - ranged competing dispersion potentials .
we demonstrate that for a suitable set of colloid - fluid , slabs - fluid , and fluid - fluid coupling parameters the competition between the effects due to the coatings and the core regions of the objects involved result , when one changes the temperature @xmath1 , the chemical potential of the fluid @xmath2 , or the distance @xmath0 between the objects , in _ sign change _ of the casimir force acting between the colloid and the slab , as well as between the colloids .
furthermore , we show that the same can also be achieved for the ntf between these objects .
the last can be used for governing the behavior of objects , say colloidal particles , at small distances , say in colloid suspensions for preventing flocculation .
it can also provide a strategy for solving problems with handling , feeding , trapping and fixing of microparts in nanotechnology .
data for specific substances in support of the experimental feasibility of the theoretically predicted behavior of the ntf have been also presented . |
the travelling salesman problem ( tsp ) is to find a least cost hamilton cycle in an edge weighted graph .
it is of the most widely studied combinatorial optimization problems and is well - known to be np - hard .
the tsp model has been used to formulate a wide variety of applications . for details
we refer the reader to the well - known books @xcite as well as the following papers @xcite . for some applications ,
more than linear combinations of distances between consecutive nodes are desirable in formulating an objective function .
consider the problem of determining an optimal routing of a unmanned aerial vehicle ( uav ) which has a list of targets at specific locations .
this can be modelled as a tsp which requires a tour that minimizes the distance travelled .
however , such a model this neglects to factor in the physical limitations of the vehicle , such as turn radius or momentum . to illustrate this idea , in figure [ fig : figure1 ] we give a hamilton path , in figure [ fig : figure2 ] we give the corresponding flight path , and figure [ fig : figure3 ] shows a route which is longer but
can be travelled at a greater speed and hence reducing the overall travel time . to model the traversal time
, we can introduce penalties for pairs of ( not necessarily adjacent ) edges to force a smooth curve for its traversal . in this paper
we consider a generalization of the tsp which can be used to model similar situations and contains many variations of the tsp , such as the angle - metric tsp @xcite , dubins tsp @xcite and the tsp @xcite as special cases .
let @xmath2 be an undirected graph on the node set @xmath3 with the convention that all indices used hereafter are taken modulo @xmath4 .
for each edge @xmath5 a nonnegative cost @xmath6 is given .
a _ cycle _ in @xmath7 is a subset of @xmath8 that forms a path starting and ending at the same node .
note that a cycle can be represented by either an ordered set of nodes or a set of edges in @xmath7 .
a _ hamilton cycle _ or _ tour _ in @xmath7 is a cycle containing every node of @xmath7 .
let @xmath9 be a tour in @xmath7 and let @xmath10 be the set of all tours in @xmath7 .
the edges @xmath11 and @xmath12 , @xmath13 are _
@xmath0-neighbours _ on @xmath14 if and only if a shortest path between @xmath15 and @xmath16 on @xmath14 containing these edges has exactly @xmath0 edges , for @xmath17 . here
the shortest path refers to the path with the least number of edges , rather than the minimum cost path .
thus @xmath15 and @xmath16 are 2-neighbours in @xmath14 if and only if they share a common node in @xmath14 .
let @xmath18 be the cost of the pair @xmath19 of edges and @xmath20 are @xmath21-neighbours on @xmath22 .
assume that @xmath23 for every pair of edges @xmath24 . then the _ @xmath0-neighbour tsp _ ( tsp(@xmath0 ) )
is defined as in @xcite @xmath25 a closely related problem , the quadratic tsp ( qtsp ) , is defined as follows : @xmath26 where @xmath27 .
note : @xmath28 thus when @xmath29 ( for @xmath4 even ) or @xmath30 ( for @xmath4 odd ) , the @xmath0-neighbour tsp reduces to the quadratic tsp .
we note tsp(@xmath31 ) is simply the original tsp . elsewhere in the literature ( e.g. @xcite , @xcite ) , the term qtsp
is sometimes used for what we refer to as tsp(@xmath32 ) .
that is , quadratic terms are allowed , but only for pairs of edges that share a node .
the bottleneck version of tsp(@xmath0 ) was introduced by arkin et al . in @xcite , denoted as the @xmath0-neighbour maximum scatter tsp .
jger and molitor @xcite encountered tsp(2 ) while studying the permuted variable length markov model .
several heuristics are proposed and compared in @xcite as well as a branch and bound algorithm for tsp(2 ) in @xcite . a column generation approach to solve tsp(@xmath32 )
is given in @xcite and polyhedral results were reported by fischer and helmberg @xcite and fischer @xcite .
the @xmath0-neighbour tsp is also related to the @xmath0-peripatetic salesman problem @xcite and the watchman problem @xcite . to the best of our knowledge ,
no other works in the literature address tsp(@xmath0 ) . referring to the uav example discussed earlier ,
it is clear that the flight subpaths depend on both the angle and distances between successive nodes . by precalculating these and assigning costs to @xmath18 for some @xmath33
, we see that qtsp is a natural model for this problem .
in fact , the flight paths may be affected by edges further downstream .
thus we can get successively better models by considering tsp(@xmath31 ) , tsp(@xmath32 ) , ... , tsp(@xmath0 ) in turn . in practice
we expect diminishing returns to take hold quickly . in this paper
we show that qtsp is strongly np - hard even if the costs are restricted to 0 - 1 values and the underlying graph is halin . in contrast ,
tsp on a halin graph can be solved in @xmath34 time @xcite .
interestingly , we show that tsp(@xmath32 ) and tsp(@xmath35 ) can also be solved on a halin graph in @xmath34 time . in fact , our approach can be extended to any fixed @xmath0 .
we note that while halin graphs have treewidth 3 , the results on graphs with bounded treewidth ( e.g. @xcite ) can not easily extended to graph problems with quadratic objective functions .
the paper is organized as follows . in section [ halinsec ]
we introduce some preliminary results and notations for the problem .
the complexity result for qtsp on halin graphs is given in section [ complexsec ] .
an @xmath34 algorithm to solve tsp(@xmath35 ) on halin graphs is given in section [ tspkhalinsec ] , which can be extended to give an @xmath36 algorithm for tsp(@xmath0 ) .
further extensions of this result to fully reducible classes of graphs is discussed in section [ reduciblesec ] .
an earlier version of the np - completeness results presented here were included as part of the m.sc .
thesis of the first author @xcite .
a halin graph @xmath37 is obtained by embedding a tree with no nodes of degree two in the plane and connecting the leaf nodes of @xmath38 in a cycle @xmath39 so that the resulting graph remains planar .
the non - leaf nodes belonging to @xmath38 are referred to as _ tree _ or _
internal _ nodes and the nodes in @xmath39 are referred to as _ cycle _ or _
outer _ nodes of @xmath40 .
a halin graph with exactly one internal node is called a _
wheel_. if @xmath40 has at least two internal nodes and @xmath41 is an internal node of @xmath38 which is adjacent to exactly one other internal node , then @xmath41 is adjacent to a set of consecutive nodes of @xmath39 , which we denote by @xmath42 .
note that @xmath43 .
the subgraph of @xmath40 induced by @xmath44 is referred to as a _ fan _ , and we call @xmath41 the centre of the fan
. see figure [ fig : fan ] . with 3 fans .
@xmath41 is the centre of fan @xmath45 . ]
@xcite [ notwheel2fans ] every halin graph which is not a wheel has at least two fans .
let @xmath2 be a graph and let @xmath46 be a connected subgraph of @xmath7 .
let @xmath47 be the cutset of @xmath48 , that is , the smallest set of edges whose removal disconnects @xmath48 from @xmath7 .
let @xmath49 be the graph obtained by contracting @xmath48 to a single node , called a ` pseudo - node ' and denoted by @xmath50 @xcite .
the edges in @xmath49 are obtained as follows : 1 .
an edge with both ends in @xmath48 is deleted ; 2 .
an edge with both ends in @xmath51 remains unchanged ; 3 .
an edge @xmath52 with @xmath53 , @xmath54 is replaced by the edge @xmath55 .
@xcite [ contracthalin ] if @xmath45 is a fan in a halin graph @xmath40 , then @xmath56 is a halin graph .
note that each time a fan is contracted using the graph operation @xmath56 , the number of non - leaf nodes of the underlying tree is reduced by one . that is , after fewer than @xmath57 fan contractions , a halin graph will be reduced to a wheel .
let @xmath41 be the centre of a fan @xmath45 , and label the outer nodes in @xmath45 in the order they appear in @xmath39 as , @xmath58 ( @xmath59 ) .
let @xmath60 be the @xmath35-edge cutset @xmath61 which disconnect @xmath45 from @xmath7 such that @xmath62 is adjacent to @xmath63 , @xmath0 is adjacent to @xmath41 but not adjacent to @xmath64 for any @xmath65 , @xmath66 and @xmath67 is adjacent to @xmath68 ( see figure [ fig : halin2 ] , @xmath69 ) .
note that any hamilton cycle @xmath14 in @xmath40 contains exactly two edges of @xmath70 .
the pair of edges chosen gives us a small number of possibilities for traversing @xmath45 in a tour @xmath14 .
for example , if @xmath14 uses @xmath0 and @xmath67 , it contains the subsequence @xmath71 ( call this a _ left - traversal _ of @xmath45 ) , if @xmath14 uses @xmath62 and @xmath0 it contains the subsequence @xmath72 ( call this a _ right - traversal _ of @xmath45 ) and if @xmath14 uses @xmath62 and @xmath67 , it contains a subsequence of the form @xmath73 , for some @xmath74 as it must detour through the centre of @xmath45 ( call this a _ centre - traversal _ of @xmath45 ) . containing fan @xmath45 .
@xmath60 is a 3-edge cutset which disconnects @xmath45 from @xmath40 . ]
many optimization problems that are np - hard on a general graph are solvable in polynomial time on a halin graph @xcite . unlike these special cases ,
we show that qtsp is strongly np - hard on halin graphs .
the decision version of qtsp on a halin graph , denoted by rqtsp , can be stated as follows : `` given a halin graph @xmath40 and a constant @xmath75 , does there exist a tour @xmath14 in @xmath40 such that @xmath76 ? '' [ t1 ] @xcite rqtsp is strongly np - complete even if the values @xmath77 and @xmath78 for @xmath79 .
rqtsp is clearly in np .
we now show that the 3-sat problem can be reduced to rqtsp .
the 3-sat problem can be stated as follows : `` given a boolean formula @xmath80 in conjunctive normal form ( cnf ) containing a finite number of clauses @xmath81 on variables @xmath82 such that each clause contains exactly three literals ( @xmath83 where for each @xmath65 , @xmath84 or @xmath85 for some @xmath86 ) , does there exists a truth assignment such that @xmath80 yields a value ` true ' ? '' from a given instance of 3-sat , we will construct an instance of rqtsp .
the basic building block of our construction is a 4-fan gadget obtained as follows .
embed a star on 5 nodes with center @xmath87 and two specified nodes @xmath88 and @xmath89 on the plane and add a path @xmath90 from @xmath88 to @xmath89 covering each of the pendant nodes so that the resulting graph is planar ( see figure [ fig : gadget ] ) .
call this special graph a 4-fan gadget .
the nodes on path @xmath90 of this gadget are called _
outer nodes _ and edges on @xmath90 are called _ outer edges_.
let @xmath91 be edges with distinct end points in @xmath90 .
note that any @xmath88-@xmath89 hamilton path of the gadget must contain all the outer edges except one which is skipped to detour through @xmath87 .
we will refer to an @xmath88-@xmath89 hamilton path in a @xmath92-fan gadget as a _ center - traversal _ as before .
we will construct a halin graph @xmath40 using one copy of the gadget for each clause and let @xmath91 correspond to literals contained in that clause .
we will assign costs to pairs of edges such that every hamilton cycle with cost 0 must contain a centre - traversal for each clause . to relate a hamilton cycle to a truth assignment , a centre - traversal which does not contain edge @xmath93 corresponds to an assignment of a _ true _ value to literal @xmath94 .
now construct @xmath40 as follows .
for each clause @xmath95 , create a copy of the @xmath92-fan gadget .
the @xmath96 , and @xmath87 nodes of the 4-fan gadget corresponding to the clause @xmath97 are denoted by @xmath98 and @xmath99 respectively .
connect the node @xmath100 to the node @xmath101 , @xmath102 .
introduce nodes @xmath103 and @xmath104 and the edges @xmath105 . also introduce a new node @xmath41 and connect it to @xmath106 and @xmath99 for @xmath107 .
the resulting graph is the required halin graph @xmath40 .
see figure [ fig : rqtsp ] .
assign the cost @xmath108 for every edge in @xmath40 .
let @xmath109 and @xmath110 .
note that every tour which contains edges @xmath111 and @xmath112 traverses every gadget using a centre - traversal .
for each gadget : assign paired cost @xmath113 for pairs of edges which are neither outer edges nor both adjacent to the same literal edge @xmath114 or @xmath115 , and for all other pairs of edges within the gadget assign cost @xmath116 . for each variable @xmath117 , @xmath118 , and all literals @xmath119 ( @xmath120 ) if @xmath121 , assign cost @xmath122 where @xmath123 and @xmath124 are edges connecting @xmath125 ( and @xmath126 ) to the respective @xmath92-fan gadget centre @xmath87 .
all other paired costs are assumed to be 0 .
suppose @xmath127 is a valid truth assignment .
then in each clause there exists at least one true literal .
consider a tour @xmath14 in @xmath40 which contains the edges @xmath111 and @xmath112 and traverses every gadget such that @xmath14 detours around exactly one literal edge which corresponds to a literal which is true in @xmath127 .
since the truth assignment is valid , such a @xmath14 exists .
clearly @xmath14 has cost 0 , since no costs are incurred by pairs of edges contained in a single gadget , nor are costs incurred of the form @xmath128 where @xmath129 .
the latter must be true because in any truth assignment , the variable corresponding to @xmath130 , say @xmath131 must be either assigned a value of true or false .
suppose a cost of 1 is incurred by @xmath128 and hence @xmath132 .
if @xmath131 is _ true _ , and @xmath133 , then @xmath134 clearly must be _ false _ , so @xmath14 can not detour to miss both @xmath135 and @xmath136 .
the same contradiction arises , if @xmath131 is _
false_. hence a _ yes _ instance of 3-sat can be used to construct a _ yes _
instance for rqtsp .
now suppose there is an optimal tour which solves rqtsp with @xmath137 .
suppose @xmath138 is such a tour .
clearly it must use edges @xmath111 and @xmath112 , and hence must traverse every gadget via a centre - traversal .
such a detour must skip a literal edge in every gadget , otherwise a cost of 1 is incurred .
suppose @xmath139 is the set of literals which are skipped .
@xmath140 for any @xmath141 , otherwise a cost of 1 is incurred .
this implies that a truth assignment which results in every literal in @xmath142 being _ true _ is a valid truth assignment to the variables @xmath143 .
that is , for each literal edge which is skipped in @xmath138 , assign _ true _ or _ false _ to the corresponding variable such that the literal evaluates to _ true _ ( if @xmath84 , set @xmath144 and if @xmath85 , set @xmath145 ) .
the truth values for any remaining variables can be assigned arbitrarily .
this truth assignment returns _ true _ for each clause since exactly one literal in each clause is detoured , and evaluates to _ true_. hence this truth assignment is a valid assignment for 3-sat .
we now examine the case when @xmath0 is fixed . note that for fixed @xmath0 , only a constant number of pairs of edges are relevant in evaluating a tour .
define edges @xmath146 and @xmath147 in a planar graph @xmath7 to be _ consecutive _ at @xmath111 in a planar embedding of @xmath7 if there exists a face which contains both @xmath15 and @xmath16 .
this may include the unbounded _
outer face_. two edges that are not consecutive are called _ non - consecutive _ edges .
[ t3 ] @xcite given a planar embedding of @xmath40 , no tour in @xmath37 contains edges @xmath146 and @xmath147 which are non - consecutive at @xmath111 .
proceed by contradiction .
suppose that there exists a planar embedding of @xmath40 and a tour @xmath14 in @xmath40 such that @xmath146 and @xmath147 are nonconsecutive edges at @xmath111 and @xmath148 . by our previous discussion regarding fan traversals ,
we only need to consider @xmath149 .
since @xmath15 and @xmath16 are non - consecutive at @xmath111 , there are edges in the planar embedding of @xmath40 , say @xmath150 and @xmath151 , such that the clockwise order of edges around @xmath111 is @xmath152 ( see figure [ fig : consechalin ] ) .
suppose @xmath38 is rooted at @xmath111 , then @xmath111 has ( at minimum ) subtrees rooted at @xmath153 and @xmath154 . since @xmath14 is a tour which contains edge @xmath15 , it must contain a path through the subtree rooted at @xmath155 from @xmath155 to @xmath156 , which we denote by @xmath157 .
similarily , it must contain a path @xmath158 in the subtree rooted at @xmath87 from @xmath87 to @xmath159 ( see figure [ fig : consechalin ] ) .
for any @xmath160 , let @xmath161 be the graph obtained from @xmath7 by deleted the vertices in @xmath45 and edges incident to a node in @xmath45 .
now we note that @xmath162 where @xmath163 has two components , one containing @xmath112 and one containing @xmath154 . restricting to @xmath14 , which also contains @xmath90 as a path
, we see that @xmath164 contains at least 2 components , which is a contradiction
. we will refer to paths that may be contained in a tour as _ candidate paths _ and _ @xmath0-candidate paths _ when the length @xmath0 is specified .
these must satisfy the condition of theorem [ t3 ] .
[ cor : kpaths ] for fixed @xmath0 , @xmath40 , given tour @xmath14 and @xmath165 , @xmath166 , there are at most @xmath167 @xmath0-candidate paths containing @xmath15 which may belong to @xmath14 . from corollary [ cor :
kpaths ] , it is clear that the number of quadratic costs which are relevant is bounded above by @xmath168 for any fixed @xmath0 . since the interior nodes of @xmath40 form a tree , it is clear that any face of @xmath40 must contain an outer edge .
moreover , the following corollary will prove useful .
[ cor : face ] if @xmath40 is embedded in the plane such that it is planar and @xmath39 defines the outer face , for any outer edge @xmath15 which is contained in the outer face and face @xmath169 , every tour which does not contain @xmath15 must contain all other edges in @xmath169 .
tsp(@xmath31 ) is the same as tsp , which is solvable in linear time on halin graphs @xcite .
tsp(@xmath32 ) can also be solved in linear time by minor modifications of the algorithm in @xcite .
to do so , one simply has to store the costs of traversing the fan ( with respect to the tsp(@xmath32 ) cost function ) as the penalties at the node resulting from the fan contraction operation .
however , for fixed @xmath170 , such minor changes appear not to be a viable option .
we now develop a linear time algorithm to handle tsp(@xmath35 ) .
the following modification is applied to the cost function to simplify the problem .
note that we can restrict our attention to the @xmath34 candidate 3-paths as a result of corollary [ cor : kpaths ] .
let @xmath171 .
for each candidate triple @xmath172 , let @xmath173 consider the problem : @xmath174 [ thm : stsp ] any optimal solution to the stsp(3 ) is also optimal for tsp(3 ) .
suppose @xmath138 is a tour in @xmath10 .
then @xmath175 hence , every tour in @xmath10 has equal cost under the simplified problem as the original tsp(3 ) objective function .
it immediately follows that any optimal solution to the simplified problem is optimal for the tsp(3 ) . in view of theorem [ thm : stsp ] , we can restrict our attention to stsp(3 ) .
we note that the approach used in cornuejols et al .
@xcite where the costs of paths are assigned to the newly created arcs by solving a linear system of equations can not be extended for any @xmath0-neighbour tsp for @xmath170 as it leads to an over - determined system of equations which may be infeasible .
instead , we extend the penalty approach used in phillips et al .
the idea here is to introduce a node - weighted version of the problem where we define a penalty function on the nodes of @xmath39 .
we iteratively contract the fans in @xmath40 , storing the values of the best subpaths which traverse each fan in the penalty function .
once we are left with a wheel , we can find the optimal tour .
we reverse the order of contraction and un - contract , maintaining optimality , until we have an optimal tour in the original graph .
let us first discuss the case where @xmath40 is a halin graph which is not a wheel . in this case
, @xmath40 will have at least two fans .
let @xmath45 be an arbitrary fan in @xmath40 with @xmath41 as the centre .
label the outer nodes in the order they appear in @xmath39 , say , @xmath58 ( @xmath59 ) .
let @xmath60 be the @xmath35-edge cutset @xmath61 which disconnects @xmath45 from @xmath40 such that @xmath62 is adjacent to @xmath63 , @xmath0 is adjacent to @xmath41 and @xmath67 is adjacent to @xmath68 .
let @xmath176 , @xmath177 and @xmath178 .
there are at least two edges not connected to @xmath45 which are consecutive with @xmath0 at @xmath111 .
the edge which follows @xmath0 in the clockwise orientation of edges at @xmath111 is denoted @xmath179 , and the other is @xmath180 .
there are exactly two edges not connected to @xmath45 and incident on @xmath181 .
these edges are labelled @xmath182 . likewise , there are exactly two edges not connected to @xmath45 and incident on @xmath183 .
these edges are labelled @xmath184 .
( see figure [ fig : fanf ] . ) without loss of generality @xmath185 are in @xmath39 and @xmath186 are in @xmath38 . to complete a fan contraction operation ,
we consider the 3 types of traversals of @xmath45 .
we define a penalty function stored at nodes of @xmath39 which contains the minimum traversal of @xmath45 of each type . for any left or right - traversal of @xmath45
, there is a single path through @xmath45 .
any tour which includes @xmath62 and @xmath0 must pass through one edge of the pair incident with @xmath62 lying outside @xmath45 together with edges @xmath187 .
similarly , any tour which includes @xmath0 and @xmath67 must pass through one edge of the pair incident with @xmath67 lying outside @xmath45 together with edges @xmath188 .
any tour which includes @xmath62 and @xmath67 must also pass through one edge in of each of the pairs of edges incident with @xmath62 ( or @xmath67 ) lying outside @xmath45 .
that is , every @xmath14 containing @xmath62 and @xmath67 must contain a path containing one subsequence from the set @xmath189 .
we refer to a centre - traversal of @xmath45 which bypasses @xmath190 as a left path , one which bypasses @xmath191 for some @xmath192 $ ] as a middle path , and one which bypasses @xmath193 as a right path .
let @xmath194 be the set of pseudo - nodes in @xmath39 .
for each @xmath195 , define @xmath196 and @xmath197 @xmath198 , let @xmath199 and let @xmath200 .
@xmath201 indicates which penalty value ( to be defined shortly ) stored at node @xmath65 is to contribute to the objective function value .
the first component of @xmath201 represents the inner structure of @xmath65 ( edges @xmath202 and @xmath203 prior to fan contraction ) , and the second component of @xmath201 , the outer ( @xmath204 to @xmath180 prior to the contraction of adjacent pseudo - nodes ) .
let @xmath205 be the vector containing @xmath201 for every @xmath195 .
let @xmath206 be the restriction of @xmath205 to the vertices in @xmath56 and augmented by @xmath207 .
let @xmath208 be the penalty that is incurred if @xmath201 occurs at @xmath65 . for @xmath209
define : @xmath210 since the penalties stored at @xmath211 depend on edges which are not incident with @xmath65 , and the dependent edges may be ` absorbed ' into adjacent pseudo - nodes so not every @xmath205 is feasible for a given @xmath14 .
that is , the inner and outer structures of adjacent pseudo - nodes in @xmath39 must agree . formally , we say that @xmath205 is feasible for @xmath14 if the following conditions are satisfied for every pseudo - node @xmath65 1 .
@xmath212 and 2 .
@xmath213 , and for every pair of consecutive pseudo - nodes @xmath214 1 .
@xmath215 , 2 .
@xmath216 , 3 .
@xmath217 and 4 .
@xmath218 .
let @xmath219 be the set of all feasible @xmath220 pairs .
we refer to @xmath220 pairs as tour pairs and now only consider feasible @xmath205 . for an example of a feasible tour pair ,
see figure [ fig : rho ] .
the problem now contains a cost for every triplet of consecutive edges in tour @xmath14 and additionally , a penalty at each outer node @xmath65 .
consider the modified 3-neighbour tsp on a halin graph defined as follows : @xmath221 note that @xmath222 can be computed in @xmath223 time by storing penalty 20-tuples at each cycle node @xmath87 .
also note that there are @xmath224 feasible @xmath205 vectors for a given @xmath14 , however , we show that the optimal @xmath220 pair can be found in @xmath34-time .
initially , set @xmath225 for all @xmath201 and @xmath209 . for fan @xmath45 in @xmath40 ,
the penalties must be updated to store the costs of traversing @xmath45 when @xmath45 is contracted to pseudo - node @xmath226 .
let @xmath227 represent the traversal of @xmath45 which contains only edges in @xmath39 .
that is , @xmath228 .
then @xmath229 represents the cost incurred by selecting the edges in @xmath227 .
assign the minimum cost of the left - traversal ( which contains @xmath230 , @xmath231 and @xmath180 ) , with inner structure @xmath232 to @xmath233 .
that is , assign the sum of the costs along the traversal together with the set of penalties on the outer nodes in @xmath45 which minimizes this cost . in the case that @xmath234 , it is not possible to have an inner structure @xmath235 , and @xmath236 .
otherwise @xmath237 similarly , assign the minimum cost of the right - traversal ( which contains @xmath238 , @xmath239 and @xmath179 ) with inner structure @xmath240 to @xmath241 . in the case that @xmath242 , it is not possible to have an inner structure @xmath80 , and @xmath243 .
otherwise @xmath244 let @xmath245 , be the centre - traversal of @xmath45 which does not contain @xmath246 .
then @xmath247 represents the cost incurred by the edges in @xmath248 .
that is , @xmath249 @xmath250 for @xmath251 , and @xmath252 assign the minimum cost of the centre - traversal which contains @xmath230 , @xmath231 and @xmath238 , @xmath253 which has inner structure @xmath254 to @xmath255 . in the case
that @xmath234 and @xmath256 , there is a single path traversing @xmath45 with inner structure @xmath235 , namely , @xmath257 so @xmath258 and when @xmath259 , we assign the cost of the minimum centre - traversal with inner structure @xmath235 .
note that this path detours some @xmath260 , @xmath261 .
@xmath262 similarly , when @xmath263 and @xmath242 , there is a single path traversing @xmath45 with inner structure @xmath80 , namely , @xmath264 so @xmath265 and when @xmath242 , we assign the cost of the minimum centre - traversal with structure @xmath80 .
note that this path detours some @xmath260 , @xmath266 .
@xmath267 now consider @xmath268 .
if @xmath269 then @xmath270 is assigned the cost of the minimum path detouring @xmath260 for @xmath271 .
if @xmath63 ( @xmath68 ) is not a pseudo - node then the centre - traversal must contain @xmath272 ( @xmath273 ) and @xmath31 ( @xmath274 ) is removed from @xmath275 from the following equation .
@xmath276 now consider @xmath277 .
this is the same as the case where @xmath268 except for the inner structure .
@xmath278 all @xmath279 which has not been assigned are not associated with a feasible @xmath280 and are assigned a value @xmath281 .
suppose @xmath37 is a halin graph which is not a wheel and @xmath45 is a fan in @xmath40 .
if @xmath282 is an optimal tour pair in @xmath40 then there exists a feasible @xmath283 in @xmath56 such that @xmath284 is optimal in @xmath56 and @xmath285 .
let @xmath194 be the set of pseudo - nodes in @xmath40 .
let @xmath226 be the pseudo - node which results from the contraction of @xmath45 , and label the vertices and edges of @xmath45 as in figure [ fig : fanf ] .
given @xmath282 , we construct @xmath284 as follows .
let @xmath286 and @xmath287 corresponding to the structure of @xmath282 around @xmath45 in @xmath40 .
that is , @xmath14 contains an @xmath288 path through @xmath45 in @xmath40 , so let @xmath289 where @xmath290 @xmath291 @xmath292 and @xmath293 then @xmath294 is feasible in @xmath56 .
using equations [ betal]-[betabm ] and noting that all new triples which contain @xmath226 have @xmath295 , we get @xmath296 it remains to show that @xmath284 is optimal in @xmath56 . towards a contradiction ,
suppose there exists a tour pair @xmath297 such that @xmath298 .
@xmath299 otherwise this contradicts the calculations of the minimums in equations [ betal]-[betabm ] . using equations [ betal]-[betabm ]
, we can expand @xmath45 , extending @xmath300 to @xmath301 in @xmath40 with @xmath302 . then @xmath303 , contradicting the optimality of @xmath282 .
hence @xmath284 is optimal in @xmath56 .
we now show that @xmath279 can be updated in @xmath304-time , by introducing a structure which allows to optimally chain together the @xmath305-values for consecutive nodes in @xmath306 .
refer to the subgraph induced by the nodes @xmath307 as _ pseudo - fan _ @xmath308 for ( @xmath309 ) .
define the penalty associated with pseudo - fan @xmath308 to be @xmath310 for @xmath311 with inner structure @xmath312 and outer structure @xmath313 .
to compute @xmath310 we use the following recursion which chains together the minimum @xmath305-values through consecutive nodes in @xmath39 while maintaining feasibility of @xmath205 . for @xmath314 , @xmath315 and @xmath316 , and @xmath317 @xmath318 @xmath319 and @xmath320
it is now possible to compute the minimum traversals of @xmath45 used in equations [ betal]-[betabm ] . for example , middle paths through @xmath45 have cost which minimizes the sum of @xmath321 and @xmath322 , such that both pairs are feasible . in the worst case ,
when all outer nodes in @xmath45 belong to @xmath194 , we must compute @xmath323 and @xmath324 , which can be done in @xmath304-time . by pre - computing these
, one can evaluate the minimum traversals of @xmath45 used in equations [ betal]-[betabm ] , in @xmath304-time and hence one can update @xmath279 in @xmath304-time .
we iteratively perform the fan contraction operation , updating costs and penalties until we are left with a wheel . the optimal tour in @xmath40 skirts the cycle @xmath39 and detours exactly once through centre @xmath41 , skipping exactly one edge of @xmath39 .
orient the cycle @xmath39 in the clockwise direction .
@xmath14 contains all edges in @xmath39 except for the skipped edge , say @xmath325 , together with the two edges which detour around @xmath326 .
define function @xmath327 for each edge @xmath328 .
let @xmath329 and @xmath330 be the tree edges adjacent to @xmath64 and @xmath331 , respectively .
define @xmath332 then , the optimal tour pair has @xmath333 suppose that we fix an edge @xmath326 in @xmath14 .
then @xmath40 can be considered to be a fan as shown in figure [ fig : wheelfan ] .
fix edge @xmath334 in @xmath14 and consider @xmath40 to be a fan @xmath335
. then the minimum tour in @xmath40 can be determined by calculating the minimum of the minimum centre - traversal of @xmath335 and the tour which bypasses @xmath334 ( using , say , fan @xmath336 ) .
this can be computed in @xmath337-time using the pseudofan technique described above .
the preceeding discussion yields the following algorithm .
halin graph @xmath40 use the wheel procedure to find an optimal tour @xmath14 in @xmath40 let @xmath45 be a fan in @xmath40 contract @xmath45 to a single node @xmath226 , using the case 1 procedure .
that is , assign penalties @xmath305 to @xmath226 , assign cost @xmath116 to all triples in @xmath56 and assign costs @xmath338 to all remaining triples which are @xmath35-neighbours in @xmath40 .
halintsp3(@xmath56 ) expand all pseudo - nodes in reverse order and update @xmath14 @xmath14 each time that a fan is contracted , the number of tree nodes is reduced by 1 , and hence the fan contraction operation is performed one less than the number of non - leaf nodes in @xmath40 . the fan contraction operation can be performed in @xmath304-time , and each time it is performed , the number of nodes in @xmath40 is reduced by @xmath339 . since the wheel procedure takes @xmath34-time , the total time for the algorithm is @xmath34 . as a result of corollary [ cor : face ] , the preceding algorithm can be extended to solve tsp(@xmath0 ) in @xmath34 time by extending the penalty functions at outer nodes to accommodate subpaths of length @xmath340 .
the method presented here may be extended to tsp(@xmath0 ) for any fixed @xmath0 , however the complexity increases by a factor of @xmath340 , which is constant for fixed @xmath0 and @xmath341 when @xmath342 .
we say that a class @xmath343 of 3-connected graphs is _ fully reducible _ if it satisfies the following : 1 . if @xmath344 has a 3-edge cutset with shores @xmath48 and @xmath345 , then both @xmath49 and @xmath346 are in @xmath343 and we call @xmath7 a _ reducible _ graph in @xmath343 ; and 2
. tsp can be solved in polynomial time for the graphs in @xmath343 that do not have non - trivial 3-edge cutsets .
we call such graphs _ irreducible_. for instance , halin graphs can be understood as graphs built up from irreducible fans connected to the remainder of the graph via @xmath35-edge cutsets
. cornuejols et al .
@xcite show that the ability to solve tsp in polynomial time on irreducible graphs in @xmath343 allows to solve tsp in polynomial time on all of @xmath343 using facts about the tsp polyhedron .
we remark that the algorithm of section @xmath92 can be used to show a similar statement for tsp(@xmath0 ) . here
we consider a graph class @xmath343 that is _ fully @xmath0-reducible _ in the sense that either it can be subdivided into irreducible graphs via @xmath35-edge cutsets , or it is irreducible and it is possible to solve the @xmath0-neighbour hamilton path problem in polynomial time .
this requires solving the following problem : @xmath347 where @xmath348 is a penalty function for the pseudo - node which depends on how tour @xmath14 traverses @xmath65 , analogous to the construction for the @xmath35-neighbour tsp of section [ ksec ] .
a graph @xmath7 from a class of fully reducible graphs .
hp(@xmath0 ) and hence mtsp(@xmath0 ) can be solved in polynomial time on @xmath7 .
let @xmath14 be the minimum cost tour in @xmath7 .
let @xmath48 be an irreducible subgraph in @xmath7 .
contract @xmath48 to a single node @xmath50 , assigning penalties @xmath348 to @xmath50 . call the resulting graph @xmath49 .
assign cost @xmath116 to all @xmath0-tuples in @xmath49 which correspond to @xmath0-paths containing @xmath50 as an interior node .
all other costs remain the same in @xmath49 as in @xmath7 .
mtsp(@xmath0)(@xmath49 ) expand all pseudo - nodes in reverse order and update @xmath14 @xmath14 we recursively perform the contraction operation on the irreducible subgraphs of @xmath7 , storing the necessary tour information in the penalty at the resulting pseudo - node .
for any fixed @xmath0 , this requires a polynomial number of penalties and the least cost traversals of @xmath48 can be computed in polynomial time using a generalization of the pseudo - fan strategy above .
suppose the contraction operation is performed on a subgraph of size @xmath89 in time @xmath349 , where @xmath350 is a polynomial in @xmath89 .
each time this operation is performed , the number of nodes in the graph is reduced by @xmath89 .
this operation is performed at most @xmath4 times and it follows that the entire algorithm can be performed in polynomial time .
in this paper , we have shown that qtsp is np - hard even when the costs are restricted to taking 0 - 1 values on halin graphs .
we have presented a polynomial time algorithm to solve a restriction of qtsp , denoted tsp(@xmath0 ) on any fully reducible graph for any fixed @xmath0 . to illustrate this
, we have given an algorithm which solves tsp(3 ) on a halin graph in @xmath34 time .
it may be noted that the approach that is presented can directly be extended to the corresponding bottleneck versions on any fully reducible graph class by replacing the ` sum ' operations with ` max ' with the same time complexity .
h. bodlaender .
dynamic programming on graphs with bounded treewidth . in _
automata , languages and programming ( tampere , 1988 ) _ , volume 317 of _ lecture notes in comput .
_ , pages 105118 .
springer , berlin , 1988 .
a. fischer , f. fischer , g. jger , j. keilwagen , p. molitor , and i. grosse .
exact algorithms and heuristics for the quadratic traveling salesman problem with an application in bioinformatics .
, 166(0):97114 , 2014 .
g. jger and p. molitor .
algorithms and experimental study for the traveling salesman problem of second order . in boting yang , ding - zhu du , and caoan wang , editors , _ combinatorial optimization and applications _
, volume 5165 of _ lecture notes in computer science _ , pages 211224 .
springer berlin , 2008 . | the quadratic travelling salesman problem ( qtsp ) is to find a least cost hamilton cycle in an edge - weighted graph , where costs are defined on all pairs of edges contained in the hamilton cycle .
this is a more general version than the commonly studied qtsp which only considers pairs of adjacent edges .
we define a restricted version of qtsp , the @xmath0-neighbour tsp ( tsp(@xmath0 ) ) , and give a linear time algorithm to solve tsp(@xmath0 ) on a halin graph for @xmath1 .
this algorithm can be extended to solve tsp(@xmath0 ) on any fully reducible class of graphs for any fixed @xmath0 in polynomial time .
this result generalizes corresponding results for the standard tsp .
tsp(@xmath0 ) can be used to model various machine scheduling problems as well as an optimal routing problem for unmanned aerial vehicles ( uavs ) . |
let @xmath3 be a compact riemann surface of genus @xmath4 , and @xmath74 its canonical bundle
. for @xmath75 a complex reductive lie group with lie algebra @xmath76 , from @xcite one has the following definition : [ gc - higgs ] a @xmath75-higgs bundle on @xmath3 is given by a pair @xmath6 for @xmath7 a principal @xmath75-bundle on @xmath3 , and @xmath8 a holomorphic section of @xmath77 , for @xmath78 the adjoint bundle associated to @xmath7 .
[ def complex ] for @xmath79 one recovers classical higgs bundles as introduced in @xcite . for matrix groups ,
the definition can be reformulated in terms of vector bundles rather than principal bundles .
in particular , * an @xmath80-higgs bundle is a pair @xmath42 where @xmath55 is a rank @xmath81 holomorphic bundle on @xmath3 with fixed trivial determinant , and @xmath8 a traceless holomorphic section of @xmath82 ; * an @xmath83-higgs bundle is a pair @xmath42 where @xmath55 is a rank @xmath81 holomorphic bundle on @xmath3 with an orthogonal structure @xmath84 and a compatible trivialization of its determinant bundle , and @xmath8 is a holomorphic section of @xmath82 satisfying @xmath85 .
the construction of @xmath5-higgs bundles for a real form @xmath5 of @xmath75 goes back to @xcite ( see also @xcite , for further details ) .
the definition requires a choice of a maximal compact subgroup of @xmath86 , and the cartan decomposition @xmath87 for @xmath88 the lie algebra of @xmath89 , and @xmath90 its orthogonal complement .
note that the lie algebras satisfy the symmetric space relations @xmath91\subset\mathfrak{h}$ ] , @xmath92\subset\mathfrak{m}$ ] , and @xmath93\subset \mathfrak{h}\nonumber $ ] .
hence there is an induced isotropy representation given by @xmath94 [ real form ] given @xmath5 a real form of a complex lie group @xmath75 , a @xmath5_-higgs bundle _ is a pair @xmath6 where @xmath7 is a holomorphic principal @xmath95-bundle on @xmath3 , and @xmath8 is a holomorphic section of @xmath96 .
[ defrealhiggs ] alternatively , as done in @xcite , we may regard real higgs bundles as classical higgs bundles @xmath42 , with extra conditions reflecting the structure of the real group and its isotropy representation .
in this paper we shall mainly consider @xmath97 and @xmath98-higgs bundles , for which we recall their main properties in sections [ spectral_slr ] and [ real so ] ( for further references , see @xcite ) .
following @xcite , a classical higgs bundle @xmath42 is said to be ( semi)-stable if all subbundles @xmath99 such that @xmath100 satisfy @xmath101 .
moreover , the pair is said to be poly - stable if it can be written as a direct sum of stable higgs bundles @xmath27 for which @xmath102 .
the notion of stability can be extended to @xmath75-higgs bundles as in @xcite , as well as to @xmath5-higgs bundles ( e.g. see @xcite ) , and used to construct the corresponding moduli spaces .
as explained below in remark [ allstable ] stability considerations will not play a role in this paper .
we shall denote by @xmath103 and @xmath104 the moduli space of @xmath75-higgs bundles and @xmath104-higgs bundles , respectively .
given a homogenous basis @xmath105 of invariant polynomials for the lie algebra of a complex lie group @xmath75 , let @xmath106 be their degrees . then , the hitchin fibration ( introduced in @xcite ) is defined as @xmath107 a point in the hitchin base @xmath108 determines a section of the line bundle @xmath109 defined by @xmath110 , where @xmath111 the tautological section of @xmath112 .
the zero locus of this section is the so - called spectral curve @xmath17 associated to the higgs bundle .
the curve @xmath17 lies the total space of @xmath9 and the projection @xmath113 is a @xmath114-fold cover of @xmath3 . by considering vector bundles on it ( or its desingularization ) , one can give a geometric description of the fibres of the hitchin fibration @xcite .
in the case of classical higgs bundles the generic fibres are given by the jacobian varieties of the spectral curves @xmath17 , on which @xmath8 has a single - valued eigenvalue @xmath111 .
indeed , given a line bundle @xmath115 on @xmath17 one may recover the corresponding classical higgs bundle @xmath42 by taking @xmath116 and @xmath8 the direct image of @xmath117 . in section [ spectral_higgs ]
we describe more fully the spectral data for the specific cases of interest in this paper . by considering the embedding of @xmath118 in @xmath119
, one may identify the @xmath5-higgs bundles as points in the hitchin fibration satisfying additional constraints ( see @xcite for classical complex lie groups , and @xcite and references therein for real higgs bundles ) .
in particular , in the case of the split real form of @xmath75 , the line bundles are the torsion two points in the jacobian ( see ( * ? ? ?
* theorem 4.12 ) ) .
spectral curves are defined for all higgs bundles represented by points in the moduli spaces @xmath118 , but they are not necessarily smooth .
it follows however from bertini s theorem that on the generic fiber of the hitchin fibration the spectral curves are smooth . throughout the paper we shall consider higgs bundles over the smooth loci of the hitchin fibration , i.e. points defining smooth spectral curves , since we will further restrict our attention to those curves for which the most generic type of ramification behaviour occurs . [ allstable ]
note that if the spectral curve for a higgs bundle is smooth then the characteristic polynomial for the higgs field is irreducible .
the higgs bundle thus has no invariant subbundles and is therefore automatically stable .
hence , whilst stability is needed to define the moduli spaces , it will play no role in our discussions and we shall omit any further comment on it .
we shall recall here how to study the fibres of the hitchin fibration through spectral data , which we shall do by reviewing the methods introduced in @xcite for complex higgs bundles , and in @xcite for real higgs bundles .
since our main focus is on @xmath120 and @xmath121-higgs bundles , we shall restrict attention to those groups and their split real forms .
the spectral curve @xmath122 associated to an @xmath123-higgs bundle @xmath42 has equation @xmath124 here @xmath125 , and @xmath111 is the tautological section of @xmath112 .
the @xmath126-tuple @xmath127 defines a point in the base of the hitchin fibration ( see @xcite ) . over a dense open set in the base of the hitchin fibration ,
the spectral curve @xmath17 is smooth and the fiber can be identified ( biholomorphically ) with the prym variety @xmath128 , the subset of @xmath129 of line bundles whose direct image sheaf has trivial determinant .
the relation between @xmath130 and @xmath55 ( once a choice of @xmath131 is made ) is then @xmath132 moreover , @xmath133 , where @xmath134 denotes the norm map defined by the projection @xmath135 , the trivialization of @xmath136 determined by the trivialization of @xmath137 .
note that in the case of two - fold covers , a line bundle is in the prym variety if and only if its dual is isomorphic to the pull back by the involution switching the sheets of the cover . from definition [ real form ] , higgs bundles with structure group @xmath138
are given by classical higgs bundles @xmath42 together with an oriented orthogonal structure on @xmath55 , with respect to which the higgs field is traceless and symmetric .
moreover , by ( * ? ? ?
* theorem 4.12 ) one has that the intersection of the moduli space @xmath139 with the smooth fibres of the @xmath80 hitchin fibration is given by line bundles @xmath140 such that @xmath141 . following @xcite and @xcite , a torsion 2 line bundle @xmath115 induces an @xmath138-higgs bundle @xmath42 with @xmath142 and @xmath8 the push down of the tautological section @xmath111 .
moreover , the orthogonal structure on @xmath55 comes from an @xmath143 structure on @xmath115 . from the characteristic polynomial of an @xmath120-higgs field @xmath8 one obtains a @xmath144-fold cover @xmath113 whose equation is given by @xmath145 where @xmath111 is as in , the sections @xmath146 , and @xmath147 is the pfaffian of @xmath8 .
this curve has a natural involution @xmath148 , and is singular at @xmath149 , i.e. , at the fixed points of the @xmath150 .
the normalization of @xmath17 , which we denote by @xmath151 , is what we shall refer to as the spectral curve . as in the previous cases
, @xmath152 is generically smooth , and the involution @xmath150 extends to an involution @xmath153 on @xmath152 which does not have fixed points .
the generic fibers of the hitchin fibration can be identified with an abelian variety defined by a connected component of @xmath154 , i.e. the kernel of the norm map @xmath155 .
given a line bundle @xmath156 and a choice of @xmath131 , the vector bundle @xmath55 is recovered as @xmath157 in this case , the orthogonal structure on @xmath55 comes from the isomorphism @xmath158 ( by virtue of which @xmath159 ) .
the spectral data associated to an @xmath160-higgs bundle is defined on the desingularization @xmath152 of @xmath17 . from eq . , each pair of points @xmath161 and @xmath162 define s the same curve @xmath17 and desingularization @xmath152 .
the prym variety @xmath163 has two connected components .
moreover , from ( * ? ? ?
* lemma 1 ) all line bundles @xmath115 on @xmath164 which satisfy @xmath165 are of the form @xmath166 for some line bundle @xmath167 on @xmath164 of degree @xmath168 or @xmath169 .
the two connected components of @xmath163 correspond to the two possibilities for the parity of the degree @xmath170 .
this can also be seen as a reflection of the fact that @xmath171 has two components corresponding to the possible values for the second stieffel - whitney class of an @xmath11-bundle .
we say that the spectral data is given by @xmath172 for @xmath164 the normalized curve and @xmath173 .
[ ramdiv ] for any @xmath174-fold ramified cover
@xmath175 , the ramification divisor @xmath176 in @xmath17 relates the canonical bundles @xmath177 and @xmath9 of @xmath17 and @xmath3 , respectively , by @xmath178=k_s\otimes\pi^*k^*,$ ] where @xmath178 $ ] denotes the line bundle defined by the divisor @xmath176 . if @xmath17 is a spectral curve , i.e. contained in the total space of @xmath9 , then @xmath179 , and @xmath178=\pi^*k^{(m-1)}$ ] .
the relation between @xmath115 and @xmath55 in both and can thus be given as @xmath180^{1/2})\ .\ ] ] from definition [ real form ] one has that @xmath182-higgs bundles can be viewed as @xmath11-higgs bundles of the form @xmath183 where @xmath184 are vector bundles of rank @xmath81 with orthogonal structure , say @xmath185 , and @xmath186 for @xmath187 the orthogonal transpose of @xmath56 , i.e. @xmath188 where @xmath189 denotes the dual map . by further requiring that @xmath190 , one obtains an @xmath191 pair .
in this case , from ( * ? ? ? *
theroem 4.12 ) and along the lines of section [ spectral_slr ] one has that the intersection of the moduli space @xmath192 with the smooth fibres of the @xmath83 hitchin fibration is given by @xmath193 which satisfy @xmath141 . as for classical higgs bundles ,
given a torsion two line bundle @xmath193 one obtains an @xmath191-higgs bundle @xmath42 by taking the direct image of the line bundle @xmath194 and the push forward of @xmath111 .
given the spectral data @xmath195 of an @xmath181-higgs bundle , since @xmath115 is of order two and in the prym variety of a two fold cover @xmath196 , it is invariant under the involution @xmath150 on @xmath152 .
hence its local sections decompose into invariant and anti - invariant local sections , and thus the direct image decomposes as @xmath197 , where the summands are generated by local invariant and anti - invariant sections ( see @xcite ) .
moreover , considering the @xmath81-fold cover @xmath198 the orthogonal bundles @xmath184 are recovered by taking @xmath199 for @xmath200 .
as noted in the introduction , given a fixed surface @xmath3 and a homomorphism between two lie groups @xmath201 , there is clearly an induced map @xmath202 where @xmath203 denotes the space of representations modulo conjugation .
the correspondence between surface group representations and higgs bundles thus implies a similar induced map between @xmath5-higgs bundles and @xmath204-higgs bundles . from definition
[ gc - higgs ] one sees that for an isomorphism between complex groups there is in fact an induced map from @xmath5-higgs bundles to @xmath204-higgs bundles given by @xmath205 where @xmath206 is the map defined by the derivative at the identity of the map @xmath207 , i.e. by the map on lie algebras .
moreover , if the homomorphism @xmath207 restricts to a map between real forms of @xmath5 and @xmath204 which respects the cartan decompositions of the lie algebras , then the map @xmath207 induces a map from @xmath208-higgs bundles to @xmath209-higgs bundles . from definition [ real form ] , following the notation of section [ fibration ] , the map is given by @xmath210 the maps in and ought to acquire descriptions purely in terms of spectral data , and this shall be investigated in forthcoming sections .
the relation between the spectral curves can be deduced from its relation to the eigenvalues of the higgs field . thus if the spectral curve @xmath17 for @xmath211 is defined by @xmath212 where @xmath213 are the eigenvalues of @xmath8 , then the spectral curve @xmath214 for @xmath215 is defined by an equation of the same form except with the eigenvalues replaced by the eigenvalues of @xmath216 , leading to a map @xmath217 .
alternatively one may seek a more intrinsic understanding of the map purely in terms of the information encoded in the geometry of the covering @xmath48 and the restrictions on the spectral line bundle @xmath115 .
this is our goal for the special cases of the isogenies in and .
the isogenies can be described in several ways , including from coincidences of dynkin diagrams or in terms of representations e.g. in terms of schur functors as in ( * ? ? ?
* chapter 9 - 10 ) .
for our purposes , the representation theoretic point of view is convenient , as described in the next two sections .
the group @xmath220 acts on @xmath221 by @xmath222 .
the isogeny onto @xmath0 can be seen as coming from the fact that @xmath223 .
if @xmath224 is the symplectic form on @xmath225 preserved by matrices with unit determinant , then @xmath226 , defines a symmetric , non - degenerate bilinear form on @xmath227 .
hence one has a map @xmath228 where @xmath0 is the group of ( orientation preserving ) linear maps @xmath229 preserving the form @xmath230 .
the derivative at the identity yields an isomorphism of lie algebras given by @xmath231 [ evalsadd ] if @xmath232 has eigenvalues @xmath233 , for @xmath200 , then the image @xmath234 has eigenvalues @xmath235 . in particular , if @xmath236 then @xmath237 and the eigenvalues for @xmath234 are @xmath238 . restricted to @xmath239 , the quadratic form @xmath230 has signature @xmath240 and thus the above map between complex lie groups restricts to @xmath241 where the subscript in @xmath68 denotes the connected component of the identity ( see ( * ? ? ? * section 5.2 ) for more details ) .
the map on lie algebras similarly restricts .
indeed , given @xmath242 symmetric and trace - free for @xmath200 , and fixing a basis for @xmath243 such that the orthogonal structure has the form @xmath244 , then the map in @xmath245 where @xmath246 denotes the transpose so that in this case @xmath247 .
the precise form of @xmath56 depends on the orientation chosen for @xmath243 , i.e. on the identification @xmath248 . moreover , whilst @xmath249 , the sign of @xmath250 depends on the choice of orientation .
the group @xmath251 of volume - preserving linear maps @xmath252 has a 6-dimensional representation on the exterior power @xmath253 given by @xmath254 .
taking @xmath255 and @xmath256 in the isomorphism @xmath257 and fixing an identification @xmath258 ( i.e. fixing a volume form ) , one gets a bilinear form @xmath259 on @xmath260 which is symmetric and non - degenerate .
since an element of @xmath261 preserves the volume form , it preserves a bilinear form , and thus one has a map @xmath262 where @xmath263 is the group of ( orientation preserving ) linear maps @xmath264 preserving the non - degenerate symmetric form @xmath259 .
moreover , the map gives a double cover @xmath261 of @xmath265 , thus realising @xmath261 as @xmath266 .
the derivative at the identity gives the lie algebra isomorphism @xmath267 where @xmath268 denotes the identity map on @xmath269 and the endomorphism @xmath270 is understood to be the restriction to @xmath271 .
[ eigensum ] if @xmath272 has eigenvalues @xmath273 , then as a map on @xmath274 the image @xmath275 has eigenvalues @xmath276 . restricted to @xmath277 , the quadratic form @xmath259 has signature @xmath278 and thus @xmath73 restricts to @xmath279 with a corresponding restriction of .
in particular , if @xmath280 is symmetric , trace - free , with entries @xmath281 , fixing a basis for @xmath277 such that the orthogonal structure has the form @xmath244 , the map has the form @xmath282 as in the previous case , the precise form of @xmath56 depends on the orientation chosen for @xmath277 , i.e. on the identification @xmath283 , and a standard choice yields @xmath284 in all cases , the pfaffian is given by @xmath285 , where the sign depends on the choice of orientation .
[ notsimple ] fixing a maximal compact subgroup @xmath286 , one obtains extra structure related to the presence of two inequivalent normal @xmath287 subgroups , by virtue of which @xmath47 fails to be simple .
these subgroups have an important influence on the map induced by @xmath73 on @xmath46-higgs bundles , described in more detail in [ higgs33 ] , and considered also in [ inv33 ] .
in this section we describe the map corresponding to the isogeny @xmath288 in [ lie2 ] in terms of spectral data . after exploring the maps for the complex groups we examine the extra conditions required to understand the corresponding maps for the split real forms .
an @xmath289-higgs bundle on @xmath3 is defined by a pair of @xmath21-higgs bundles @xmath27 , for @xmath200 .
an @xmath39-higgs bundle can be viewed as a pair of this type in which the bundles have oriented orthogonal structures and the higgs fields are traceless and symmetric with respect to the orthogonal structures . up to isomorphism
, we may thus assume in this case that the bundles are of the form @xmath290 where @xmath291 is a line bundle of non - negative degree , the orthogonal structure is defined by the isomorphism @xmath292 and the higgs field is of the form @xmath293 the characteristic polynomials for the higgs fields define two spectral curves @xmath294 in the total space of @xmath9 as in eq . .
these are 2-fold covers of the riemann surface , with equations @xmath295 for @xmath296 and @xmath111 the tautological section of @xmath112 .
generically the quadratic differentials @xmath297 have simple zeros , and thus by bertini s theorem the curves @xmath19 are generically smooth . as seen in section [ spectral_slr ] , the spectral data associated to these @xmath21-higgs bundles is completed by line bundles @xmath298 . in the case of @xmath39-higgs bundles , by ( * ? ? ?
* theorem 4.12 ) the line bundles are required to satisfy @xmath299 .
note that , since @xmath298 if and only if @xmath300 , the conditions that @xmath298 and @xmath301 are equivalent to the conditions that @xmath302 and @xmath303 .
[ remtoledo1 ] the @xmath304-higgs bundles @xmath27 have associated an integer invariant known as the toledo invariant .
this can be defined in several equivalent ways , including as the degree of the line bundle @xmath291 or the euler number of the @xmath305-principal bundle associated to @xmath306 .
it can also be seen in the spectral data where it is detected by the action of the involution on the fibers of the line bundle at fixed points ( see @xcite ) .
however it is defined , the toledo invariant is bound by a so - called milnor - wood inequality . an @xmath307-higgs bundle can be described as an @xmath0-higgs bundle @xmath42 where @xmath308 decomposes as the sum of two rank 2 holomorphic oriented orthogonal bundles , and @xmath309 for @xmath310 where @xmath187 is the dual map and @xmath185 is the orthogonal structure of @xmath184 .
furthermore , higgs bundles with structure group @xmath68 have isomorphisms @xmath311 . as in the previous section
, we may take the rank two orthogonal bundles to be of the form @xmath312 and @xmath313 , for @xmath314 line bundles on @xmath3 with @xmath315 , and with orthogonal structure as in .
as in eq . , the higgs field @xmath8 in an @xmath0-higgs bundle defines a 4-fold cover @xmath316 with equation @xmath317 where @xmath318 is the pfaffian , @xmath319 , and @xmath111 the tautological section of @xmath112 .
the spectral data for an @xmath307-higgs bundle is given by the spectral data @xmath320 of the corresponding complex @xmath321-higgs bundle for which @xmath322 ( * ? ? ? * theorem 4.12 ) .
moreover , as in section [ real so ] , the direct image of @xmath115 in @xmath323 defines two line bundles which induce @xmath184 on @xmath3 .
[ iso22higgs ] from section [ lie2 ] ( see @xcite for a detailed study of this ) , the map on @xmath72- higgs bundles induced by @xmath71 is @xmath328 where the orthogonal structure on @xmath29 is the tensor product of the symplectic structures @xmath329 on @xmath330 and @xmath331 ( recall that @xmath223 ) . [ detiso]notice that the the isomorphisms @xmath332 do not uniquely determine a trivialization @xmath333 compatible with the orthogonal structure on @xmath29 . indeed , if @xmath334 are local oriented frames for @xmath30 satisfying @xmath335 , then both @xmath336 and @xmath337 are orthonormal local frames for @xmath29 but they have opposite orientations .
they determine the two inequivalent choices for @xmath333 .
following the notation of section [ specsl2 ] , the map @xmath71 can then be put in the form @xmath338 we shall first construct the spectral data @xmath339 associated to the @xmath340-higgs bundle obtained via @xmath71 , and then specialize to the spectral data associated to the split real forms @xmath341 and @xmath68 . [ propso21 ] let @xmath342 be the spectral data corresponding to an @xmath289-higgs bundle , where each @xmath19 is defined by @xmath343 for @xmath296 , and @xmath344 .
then the pair @xmath339 given by * @xmath35 is the fiber product curve , and * @xmath345 is the line bundle as in diagram , gives the spectral data associated to an @xmath321-higgs bundle .
the total space of @xmath346 can be identified with the fiber product @xmath347 .
the fiber product @xmath23 may thus be regarded as a subvariety of either @xmath348 or of the total space of @xmath346 .
the fiber product @xmath349 , as a curve in the total space of @xmath346 , is defined by the conditions @xmath350 where @xmath351 denotes the tautological section on @xmath346 .
it follows that for a generic choice of @xmath352 this curve is smooth .
moreover , if @xmath353 denotes the image of @xmath349 under the map @xmath354 given by fiberwise addition , then it is defined by the conditions @xmath355 hence , the four - fold cover @xmath356 is defined by the equation @xmath357 which from [ real so ] and eq . , is the spectral curve of an @xmath321-higgs bundle .
the curve @xmath356 is generically singular , with singularities over the zeros of @xmath358 , and by construction the the map @xmath359 is an isomorphism on the smooth locus of @xmath356 .
the involution @xmath148 which preserves @xmath360 and @xmath361 induces an involution @xmath362 on @xmath363 . when needed we shall denote the involution on @xmath364 by @xmath365 , for @xmath200 and on @xmath366 by @xmath367
. the fixed points of @xmath365 are the zeros of @xmath297 , and thus , since the zeros of @xmath368 and @xmath369 are generically different , generically @xmath370 does not have any fixed points .
it is clear from that @xmath367 descends to the involution @xmath371 on the singular curve @xmath356 , where it has fixed points at the branch locus . in order to see that @xmath372 is the spectral line bundle associated to an @xmath321-higgs bundle one has to show that @xmath373 .
since @xmath374 one has that @xmath375 and so the line bundle @xmath376 is sent to its dual by the involution @xmath367 .
hence , the line bundle @xmath372 on @xmath349 is in @xmath377 as required .
[ i2spec ] the spectral data @xmath378 induced by an @xmath289-higgs bundle @xmath379 , as in proposition [ propso21 ] corresponds to the spectral data of the @xmath321-higgs bundle @xmath380 $ ] . as seen in proposition [ propso21 ]
the spectral curve @xmath349 is the curve associated to an @xmath321-higgs bundle .
furthermore , from eq .
the curve is indeed the one associated to the higgs bundle in the image of @xmath379 through @xmath71 . with the notation of below , in order to corroborate that the line bundle @xmath372 is indued the spectral line bundle of the image @xmath321-higgs bundle , note that for any line bundles @xmath381 on @xmath19 one has @xmath382 applying this to the line bundles @xmath20 and using the relation with @xmath383 , we get @xmath384 whereas by the vector bundle on @xmath3 defined by @xmath24 is @xmath385 recall that @xmath386 corresponds to the ramification divisor @xmath387 , while the ramification divisors @xmath388 satisfy @xmath389=\pi_i^*k$ ] .
it follows that @xmath390=p_1^*[r_1]\otimes p_2^*[r_2 ] = ( \pi^*k)^2,\ ] ] and hence that @xmath391 as required .
let @xmath342 be the spectral data of an @xmath67-higgs bundles , for @xmath200 .
then one has that @xmath392 , i.e. @xmath393:= \{m\in { \rm prym}(\hat s_4,\hat s_4/\hat{\sigma}_4)~|~m^2\cong \mathcal{o}\}.\end{aligned}\ ] ] thus as seen in section [ real so ] the line bundle @xmath24 defines an @xmath307-higgs bundle . since @xmath394
, it follows that @xmath395 .
this means that under the projection @xmath396 the direct image sheaf @xmath397 splits as the sum of two line bundles @xmath398 , generated by @xmath367-invariant and anti - invariant local sections .
the relation between the different covers of the riemann surface and the line bundles on them is depicted in the following diagram : @xmath399&p_*{\mathcal{l}}=\cl_+\oplus\cl_-\ar[d]\\ \hat{s}_4=s_1\times_{\sigma}s_2\ar[d]^{\pi}\ar[r]^{p}&\hat{s}_4/{\hat{\sigma}_4}\ar[dl]^{\hat \pi}\\ \sigma & } \ ] ] in particular , this implies that @xmath400 moreover , @xmath366 is the normalization of the spectral curve @xmath401 and the involution @xmath367 on @xmath366 corresponds to the involution @xmath402 on @xmath356 .
it follows that multiplication by the tautological section @xmath111 interchanges @xmath403 and @xmath404 , and thus that the higgs field on @xmath405 has the form as in .
we thus get : [ i2realhiggs]consider @xmath406 the spectral data for a point in @xmath407 represented by @xmath408 and let @xmath409 be defined as in proposition [ propso21 ] .
then @xmath409 is the spectral data for the point in @xmath324 represented by @xmath410 where @xmath56 is as in eq . .
in this section we investigate the induced map @xmath411 and its restriction to the split real forms @xmath46 and @xmath412 . from definition [ real form ] ,
an @xmath46-higgs bundle on @xmath3 is holomorphic @xmath0-principal bundle together with a symmetric higgs field .
equivalently it can be viewed as a pair @xmath413 where @xmath55 is an oriented holomorphic rank 4 orthogonal vector bundle , i.e. a vector bundle with a holomorphic symmetric non - degenerate bilinear paring @xmath84 , and a compatible isomorphism @xmath414 , and the higgs field @xmath415 is traceless and symmetric with respect to @xmath84 .
recall from section [ rec1 ] that the spectral curve for an @xmath22-higgs bundle @xmath42 is a ramified 4-fold cover @xmath113 in the total space of @xmath9 with equation @xmath416 for @xmath417 and @xmath111 the tautological section of @xmath112 . for generic choices of @xmath418
the curve is smooth and has only the most generic ramification , i.e. in fibers over the branch locus there are two unramified points and one order two ramification point .
the spectral data is completed by a line bundle @xmath140 . for an @xmath46-higgs bundle , from ( * ? ? ?
* theorem 4.12 ) the spectral line bundle @xmath140 satisfies the extra condition @xmath322 .
an @xmath419-higgs bundle can be described as @xmath1-higgs bundle @xmath42where @xmath55 decomposes as the sum of two rank 3 holomorphic oriented orthogonal bundles , say @xmath308 with orthogonal structures @xmath185 and compatible isomorphisms @xmath420 .
as seen in sections [ season]-[real so ] , for any @xmath1-higgs bundle the higgs field @xmath8 defines a 6-fold cover @xmath421 with equation @xmath422 where @xmath423 is the pfaffian and @xmath319 .
the spectral data of an @xmath263-higgs bundle is then a pair @xmath424 where @xmath61 is the desingularization of @xmath425 and @xmath426 where @xmath153 is the ( fixed - point - free ) involution inherited from @xmath425 @xcite .
when @xmath427 the spectral data corresponds to an @xmath428-higgs bundle . using the results in section [ lie1 ] for the induced action of @xmath73 on vector bundles and
lie algebras , we get the map between complex higgs bundles @xmath429 the orthogonal structure @xmath430 is induced by the combination of and the trivialization of @xmath136 . as in the case of the map defined by @xmath71 ( see remark [ detiso ] ) an isomorphism @xmath431 such that @xmath432 agrees with the trivialization of @xmath433 determined by @xmath84 , is determined only up to a choice of sign .
there are thus two ( oppositely oriented ) possible conventions for determining the @xmath1 structure on @xmath44 .
this choice plays a role in the map induced by @xmath73 on the base of the hitchin fibrations of the moduli spaces ( see section [ basemap3 ] ) .
the vector bundle @xmath55 of an @xmath46-higgs bundle @xmath42 has an oriented orthogonal structure , i.e. an associated pair @xmath434 where @xmath435 is a holomorphic orthogonal structure on @xmath55 , and @xmath436 is a compatible isomorphism @xmath437 trivializing its determinant .
the orthogonal structure induces an isomorphism ( by abuse of notation , also denoted by @xmath435 ) @xmath438 using @xmath436 as the trivialization of @xmath136 required in the construction of @xmath84 , yields an isomorphism @xmath439 which satisfies @xmath440 where @xmath441 are regarded as bilinear forms on @xmath44 , and @xmath442 . by using a local oriented orthonormal frame to compute @xmath443
, it can be seen that @xmath443 satisfies @xmath444 .
taking the @xmath53 eigenspaces of @xmath443 thus gives a decomposition @xmath445 the orthogonal structure on @xmath44 restricts to orthogonal structures on each summand , so that the structure group reduces to @xmath446 .
the structure groups of the bundles @xmath44 and @xmath447 can be reduced to @xmath47 and @xmath448 respectively .
the two copies of @xmath287 are precisely the normal subgroups mentioned in remark [ notsimple ] by virtue of which @xmath47 fails to be simple ( see @xcite ) . with respect to the reduction in , the higgs field @xmath449 has the form in , where @xmath8 and @xmath56 are related as in eq .
and @xmath450 is the orthogonal transpose . denoting orthogonal structures on @xmath52 by @xmath451 , we thus get : [ reali3 ]
the isogeny @xmath49 induces the following map between @xmath46- higgs bundles and @xmath412-higgs bundles : @xmath452 where if @xmath55 has orthogonal structure @xmath435 then the bundles @xmath52 have orthogonal structures @xmath451 , and @xmath8 and @xmath56 are related as in eq . .
the precise form of @xmath56 depends on the choices of the orientations of @xmath44 and @xmath52 .
different choices will change the sign of @xmath453 , i.e. of the pfaffian of @xmath454 .
given @xmath22-spectral data @xmath58 , with @xmath17 defined by and @xmath455 , we build @xmath1-spectral data using a construction similar to the fiber product construction in section [ iso22curve ] , except in this case we take the product of @xmath58 with itself , i.e. in diagram we have @xmath456 and @xmath457 .
the resulting curve has both singularities and additional symmetries that are absent when @xmath360 and @xmath361 are different .
our construction takes both of these features into account in an essential way .
the curve @xmath458 is a 16-fold cover of the riemann surface @xmath3 . over a generic point in the hitchin base , @xmath17 is smooth and @xmath458 has two smooth components , namely the diagonal @xmath459 and another one which we denote by @xmath460 .
the intersection of these components lies in fibers over the branch locus of the covering @xmath135 . viewing the curve @xmath461 in the total space of @xmath346 , the involution @xmath462 interchanges the copies of @xmath17 , and
thus the fixed point set of @xmath463 is @xmath464 .
the quotient map @xmath465 commutes with the projection onto @xmath3 .
it is an unramified double cover on @xmath466 but has ramification points in the fibers over the base locus of @xmath135 . using the biholomorphism @xmath467 given by @xmath468 we can view the quotient @xmath469 as a curve in the total space of @xmath470 .
we define @xmath471 and denote by @xmath472 the composition of @xmath473 and @xmath474 .
a depiction of the relation between the above curves and projections is given in diagram below . by abuse of notation , we denote by @xmath475 the restrictions to @xmath460 of the projection maps to the two factors of the full fiber product : @xmath476_{\pi}&\ar@<-.45ex>[l]_{p_1 } \ar@<.45ex>[l]^{p_2}(s\times_{\sigma}s)_0\ar[d]_{\pi_{0 } } \ar[r]^{\hat{\pi}_{\tau}}&\hat{s}_6\ar@<1ex>[ld]^{\hat{\pi}_6}\\ & \sigma & & & & } \end{aligned}\ ] ] [ uno2]for generic points in the hitchin base , the ramification divisors @xmath176 on @xmath17 , @xmath477 on @xmath460 and @xmath478 on @xmath61 for the projections @xmath479 and @xmath480 respectively , are related as follows : @xmath481 the ramification divisor for a covering @xmath482 is defined by @xmath483 where the weights @xmath484 are such that @xmath485 is the multiplicity of @xmath486 at @xmath487 . if @xmath17 is a generic spectral curve and @xmath488 is a point in the branch locus of @xmath135 then we can write @xmath489 , with @xmath490 ( with weight 1 ) but @xmath486 unramified at @xmath491
. then the fiber of @xmath469 over @xmath488 consists of points @xmath492 , [ y_1,y_2],[y_1,y_3],[y_2,y_3]\}$ ] . of these , @xmath493 $ ] and @xmath494 $ ] land in @xmath478 , each with weight 1 , while @xmath495)=(y_1,y_1)$ ] is in @xmath496 ( with weight 1 ) .
notice now that in the fiber over @xmath488 one has @xmath497)+sym([y_1,y_3]))\end{aligned}\ ] ] as required .
[ uno1 ] the curve @xmath498 is generically smooth , and the canonical desingularization of its projection to @xmath9 through the @xmath474 map , which is the spectral curve of an @xmath263-higgs bundle . in order to prove the proposition
, one needs to show the following hold : 1 .
the curve @xmath61 is generically smooth ; 2 . under the projection @xmath499 which on each fiber is given by @xmath500 , the image
@xmath501 is a spectral curve defined by an equation of the form in with @xmath502 3 .
the projection @xmath503 is an isomorphism away from the singularities of @xmath425 at its intersection with the zero section of @xmath9 . to prove the above items ,
let @xmath504 be defined by the zero locus of . as a curve in the total space of @xmath346 , the fiber product @xmath458 is defined by the conditions @xmath505 , where @xmath506 denote the tautological sections of the two summands in @xmath346 .
after using the transformation @xmath474 defined in to realize the curve as a subvariety of @xmath470 , the component @xmath61 is described locally ( i.e. with respect to a trivialization of the bundles ) as the zero locus of the map @xmath507 given by @xmath508 the projection @xmath509 onto the total space of @xmath9 is given locally by @xmath510 .
direct computation shows that away from @xmath511 , the map is a biholomorphism onto the curve define by the equation @xmath512 where @xmath111 is the local fiber coordinate on @xmath9 .
this curve , which we denote by @xmath425 , necessarily has singularities at the zeros of @xmath513 , but is otherwise smooth for generic choice of @xmath514 . for such choices , singularities in @xmath61 can occur only at points where @xmath511 .
direct computation shows that the derivative of @xmath515 has full rank at such points provided @xmath513 and @xmath516 have no common zeros , proving the proposition .
the involution @xmath402 on @xmath425 lifts to the involution @xmath517 given on fibers away from the branch locus by the map @xmath518\mapsto sym[y_k , y_l]$ ] , where @xmath519 . on a fiber with a ramification point , say @xmath520 , the involution maps @xmath521 $ ] to @xmath522 $ ] . on a regular fiber
away from the base locus , where we can write @xmath523 , the coordinate @xmath111 on @xmath425 has values @xmath524 for @xmath525 .
but @xmath526 and hence @xmath527 where @xmath519 , i.e. @xmath371 corresponds to the action of @xmath150 .
assuming that only the most generic type of ramification occurs , the computation is similar on fibers over the branch locus . to obtain the correspondence induced by @xmath73 between spectral line bundles we begin as in section [ iso22curve ] , but with @xmath457 , and consider @xmath528
since @xmath24 is invariant under @xmath463@xmath529 , its direct image on @xmath498 decomposes as a sum of rank one locally free sheaves generated by the invariant and anti - invariant local sections , i.e. @xmath530 similarly @xmath531 moreover , if @xmath532 $ ] is the line bundle on @xmath460 defined by the divisor @xmath496 , then @xmath533^{-1}$ ] ( see for example ( * ? ? ? * lemma 3.1 ) or @xcite ) .
we define @xmath534 while we have defined @xmath24 only on @xmath535 , this distinction disappears in @xmath536 .
this is a consequence of the fact that anti - invariant local sections must vanish on the fixed points of @xmath463 , so that the sheaf on @xmath537 generated by the anti - invariant sections has support only on @xmath469 .
[ landm ] since @xmath24 is clearly invariant under pullback by the involution @xmath463 and admits a lift of @xmath463 which acts as identity on fibers over all fixed points , it follows ( see for example @xcite ) that @xmath538 for some line bundle @xmath539 on @xmath61 .
we thus get @xmath540 with the summands @xmath539 and @xmath541 generated by the @xmath463-invariant and anti - invariant local sections of @xmath542 respectively . with this choice of @xmath539
we get @xmath543 [ prymmap ] if @xmath544 then @xmath545 .
let @xmath115 be defined by the divisor @xmath546 on @xmath17 given by @xmath547 for @xmath548 the divisor @xmath549 on @xmath460 , one has that @xmath548 is in the linear system of @xmath24 .
moreover , when @xmath488 is not in the branch locus of @xmath550 , so that @xmath551 has four distinct points @xmath552 , one has that @xmath553)\big):=\hat{\pi}_{\tau}^{*}\mathcal{c}_x.\ ] ] on the other hand , if @xmath488 is a branch point , then under our genericity assumptions on @xmath17 , we can assume that @xmath551 has one point ( say @xmath520 ) where @xmath550 has multiplicity two , and two points ( say @xmath491 ) where @xmath550 is unramified .
then @xmath554 with @xmath555+(d(y_2)+d(y_3))\cdot sym[y_2,y_3]+[\sum_{i=2,3}(d(y_1)+2d(y_i))\cdot sym[y_1,y_i]\ ] ] where the factor 2 in the last term comes from the fact that the projection @xmath556 is ramified at the points @xmath557 and @xmath558 , while @xmath559 is ramified at the points @xmath560 and @xmath561 .
moreover , from remark [ landm ] one has that @xmath562 and @xmath563 , and hence @xmath564 is defined by and is a divisor in the linear system of @xmath565 .
notice that on fibers over branch points of @xmath550 one has @xmath566=[y_2,y_3]$ ] .
then , denoting the branch locus of @xmath550 by @xmath567 , one gets @xmath568\
\mathrm{if}\ x\in \sigma - b ; \\ ( d(y_1)+d(y_2)+d(y_3))\cdot ( [ [ y_1,y_1]]+2[[y_1,y_2]])\ \mathrm{if}\ x\in b , \end{cases}\label{labelb}\ ] ] where @xmath569 $ ] denotes the point in @xmath570 and @xmath571 is the norm map for the covering @xmath572 .
finally , note that if @xmath573 then we can pick @xmath546 so that @xmath574 for all @xmath575 and hence @xmath576 , as required .
we have shown that the map @xmath577 sends spectral data for an @xmath22-higgs bundle to @xmath1-spectral data .
we now show that this map is compatible with the map given by .
[ rank3main ] the spectral data @xmath578 induced by an @xmath261-higgs bundle @xmath42 via and corresponds to the spectral data of the @xmath263-higgs bundle @xmath579 $ ] . by the higgs field in
@xmath579 $ ] is @xmath580 , viewded as a map on @xmath581 .
let @xmath582 and @xmath583 be the characteristic polynomials for @xmath8 and @xmath584 respectively .
then a calculation based on remark [ eigensum ] shows that the coefficients are related by .
it follows that the curve @xmath61 is the spectral curve for the @xmath263-higgs bundle @xmath579 $ ] .
moreover , proposition [ uno1 ] then shows that @xmath584 can be recovered by pushing down the section @xmath585 , where @xmath111 is tautological section of @xmath9 .
the direct image @xmath586^{1/2})$ ] acquires a special orthogonal structure in the usual way , with the trivialization of its determinant bundle determined by the prym condition @xmath587 . in order to see that @xmath586^{1/2})=\lambda^2e$ ] , consider the ( singular ) full fiber product @xmath458 and for any line bundle @xmath167 on @xmath17 let @xmath588 , where now the projections are from the entire @xmath458 . arguing as before , since @xmath589 is invariant under the involution @xmath463
, it follows that its direct image under the projection @xmath590 decomposes as the sum of rank one coherent sheaves @xmath591 .
though in principle these summands need not be locally free , their restriction to @xmath460 is well - behaved .
indeed , given the projections @xmath113 and @xmath592 one gets @xmath593 but the left- and right - hand sides of this identification are both @xmath594-eigenspace decompositions for @xmath595-actions that are compatible with the projection map .
we can thus identify @xmath596 .
notice , furthermore , that the support of @xmath597 is @xmath460 , so that @xmath598 where dropping the tilde on @xmath589 denotes the restriction to @xmath460 .
it follows that @xmath599 in particular , taking @xmath600^{1/2}$ ] where @xmath176 is the ramification divisor on @xmath17 , so that @xmath601 , it follows that gives @xmath602 .
it thus remains to show that @xmath603^{1/2}$ ] .
but by lemma [ uno2 ] and @xmath604^{1/2}\\ & = \pi_{\tau}^{*}(\mathcal{m}\otimes[\hat{r}_6]^{1/2}\otimes t^{-1}),\end{aligned}\ ] ] where @xmath539 is as in remark [ landm ] and @xmath605 is as above .
it follows that @xmath606^{1/2}\otimes t^{-1}\otimes t = \mathcal{i}_3(l)\otimes [ \hat{r}_6]^{1/2},\ ] ] as required .
we shall end this section with a discussion of the relation between our construction and the so - called trigonal construction of recillas ( see @xcite or @xcite ) . given a smooth curve @xmath3
, this construction relates a smooth four - fold cover of @xmath3 to a six - fold cover with a fixed - point - free involution , whose quotient is thus a three - fold smooth cover . taking @xmath17 as the four - fold cover , the curve @xmath607@xmath608 is precisely the corresponding six - fold cover with involution .
this is most easily seen by considering the fibers of the covering maps onto @xmath3 . at the regular fibers where ,
if the fiber of @xmath17 over a point @xmath609 consists of points @xmath610 then the fiber of @xmath469 over @xmath488 consists of the points corresponding to the unordered pairs @xmath611,[y_1,y_3],[y_1,y_4],[y_2,y_3],[y_2,y_4 ] , [ y_3,y_4]\}$ ] ( where we have dropped the dependence on @xmath488 to simplify the notation ) . considering @xmath113 a smooth cover with only the most generic ramification , if @xmath488 is in the branch locus and @xmath489 ( as in the proof of proposition [ prymmap ] ) , then the fiber of @xmath612 consists of the unordered pairs @xmath613,[y_1,y_2],[y_1,y_3],[y_2,y_3]\}$ ] .
this relation between @xmath17 and @xmath612 can be regarded as a map from @xmath17 to the third symmetric product of @xmath612 , mapping a point in @xmath17 to the unordered pairs containing that point .
this determines a correspondence defined by an effective divisor on @xmath614 , but for the sake of comparison with our construction , we use the map @xmath474 to replace @xmath612 with @xmath61 . define the effective divisor @xmath615 by @xmath616\in\mathbb{s}}\big(y , sym[y , y']\big).\ ] ] notice that @xmath617 is contained in @xmath618 and is given explicitly by @xmath619\big)\\ & + \sum_{x\in b}\bigg(\sum_{j=1}^3(y_1(x),sym[y_1(x),y_j(x)]\big)+\sum_{i=2}^3\sum_{j\ne i}\big(y_i(x),sym[y_i(x),y_j(x)]\big)\bigg).\end{aligned}\ ] ] here the set @xmath620 , as in , is the branch locus of @xmath486 , and at @xmath621 we label the points in the fiber so that @xmath486 has degree 2 at @xmath622 .
the map @xmath623\big),\end{aligned}\ ] ] is injective and its image is precisely the divisor @xmath617 . using the correspondence defined by @xmath617 we may define @xmath624&\mapsto [ nm_2(\pi_1^{*}(d)\cap\delta)]\end{aligned}\ ] ] for @xmath625 the projection @xmath626 , and the norm map @xmath627 .
the map @xmath628 restricts to a map between prym varieties , i.e. @xmath629 this map agrees with the map defined by virtue of proposition [ prymmap ] , i.e. with @xmath630 given by .
let @xmath631 be defined by divisor @xmath546 on @xmath17 . with @xmath564 as in the proof of proposition [ prymmap ] , it follows from and that @xmath632 and hence ( as in the proof of proposition [ prymmap ] ) that @xmath564 is in the linear system for @xmath63 .
it thus follows from proposition [ prymmap ] that @xmath628 defines a map between the indicated prym varieties .
we now examine the consequences of imposing the additional condition @xmath633 on the spectral data @xmath58 described in section [ specral33 ] . under the assumptions and notation of proposition [ prymmap ] , if @xmath634 , then @xmath565 is a point of order two in @xmath635 .
the map @xmath636 defined by is a group homomorphism between jacobians .
thus , in particular , if @xmath634 then @xmath637 .
it follows ( see ( * ? ? ?
* theorem 4.12 ) ) that our construction maps spectral data for @xmath46-higgs bundles to spectral data for @xmath412-higgs bundles . as in the rank two case discussed in section [ realform22 ] , if @xmath565 is a point of order two in @xmath638 then it is invariant under the involution @xmath150 . under the projection @xmath639
the direct image sheaf @xmath640 thus splits as the sum of two line bundles @xmath641 , generated by @xmath150-invariant and anti - invariant local sections . in particular , @xmath642 we get a diagram similar to : @xmath643&p_*({\mathcal{i}_3(l)})=\mathcal{i}_3(l)_+\oplus\mathcal{i}_3(l)_-\ar[d]\\ \hat{s}_6\ar[d]^{\hat{\pi}_6}\ar[r]^{p}&\hat{s}_6/{\sigma}\ar[dl]^{\pi_{\sigma}}\\ \sigma & } \ ] ] recall from proposition [ rank3main ] that @xmath61 is the normalization of the spectral curve @xmath644 and that the involution @xmath150 on @xmath61 corresponds to the involution @xmath371 on @xmath425 .
it follows that multiplication by the tautological section @xmath111 interchanges @xmath645 and @xmath646 , and thus that the higgs field on @xmath647 has the form as in .
combined with proposition [ reali3 ] , we thus get : let @xmath58 be the spectral data for a point in @xmath648 represented by a higgs bundle @xmath42 with orthogonal structure @xmath435 on @xmath55 , and isomorphism @xmath649 .
let @xmath650 be defined as and , i.e. @xmath651 and @xmath652 then @xmath650 is the spectral data for the point in @xmath653 represented by @xmath654 where the bundles have oriented orthogonal structures @xmath655 as in [ higgs33 ] , and @xmath56 is as in .
thus far we have examined the maps induced by the isogenies on individual higgs bundles or their spectral data . in this section
we collect together some remarks about the induced maps on the corresponding moduli spaces and on their hitchin fibrations .
we note that whilst the maps on higgs bundles ( given in [ iso22higgs ] and proposition [ reali3 ] ) are defined for all higgs bundles , but do not obviously preserve stability properties . on the other hand , the maps on spectral data ( see propositions [ propso21 ] and [ rank3main ] ) automatically preserve stability but apply only to generic points in the moduli spaces - where the stability condition is vacuous .
pending further work we thus limit our remarks to the dense open sets in the moduli spaces which exclude the non - generic fibers of their hitchin fibrations .
we denote these sets by @xmath656 .
note that for the groups in the isogenies studied in this paper , the special linear groups can be identified as the spin groups for the special orthogonal groups .
the induced map @xmath657 is a finite map .
as seen in @xcite , for any @xmath81 , the moduli space @xmath171 has two components corresponding to the two possible values for the second stieffel - whitney class of an @xmath11-principal bundle . in contrast , the underlying holomorphic bundles for the higgs bundles in @xmath658 and @xmath659 have just one topological type , and the moduli spaces are connected .
the images of the maps @xmath71 and @xmath73 thus see just one of the components in @xmath660 or @xmath661 .
indeed , this can be understood from the point of view of the surface group representations corresponding to the higgs bundles via non - abelian hodge theory . from this point of view
, the component in the image of the map contains precisely the representations in @xmath11 which lift to @xmath662 .
we therefore see only the higgs bundles in which the underlying holomorphic bundle has @xmath663 .
the situation is more nuanced for the restriction of the maps to the moduli spaces for the split real forms . in this case , the underlying holomorphic bundles have more complicated topology than in the case of the higgs bundles for the complex groups .
moreover , fixing the topological type of the bundle does not ensure connectedness of the components .
what remains true is that the images of the maps @xmath71 and @xmath73 contain only those components of the moduli spaces in which the higgs bundles correspond to representations which lift to the appropriate spin group . in terms of hitchin fibrations
, one should note that for each @xmath664 , the bases of the fibrations of @xmath665 and @xmath11-higgs bundles are the same . in the case of @xmath383 the base is @xmath666 , and for @xmath667 it is @xmath668 . in order to understand the maps
induced on these bases it is necessary to understand exactly the relation between coordinates of a point in the base and the coefficients in the defining equation for the spectral curve .
this is completely straightforward for @xmath669 , where the two coincide , but less so in the case of @xmath11 where the relation is complicated by the role of the pfaffian .
in fact , the maps @xmath670 ( for @xmath664 ) do not unambiguously descend to the base of the fibration .
the ambiguity stems from the fact that the induced orthogonal structures on @xmath29 or @xmath44 do not have a canonical orientation .
the choice of orientation corresponds on the one hand to a choice of trivialization on the determinant bundles , and on the other hand to a choice of sign in the pfaffian .
once the isomorphism @xmath671 is fixed , the map determines a map on the base of the hitchin fibrations for @xmath658 and @xmath660 .
the explicit form of the map depends on the generators chosen for the rings of invariant polynomials .
taking the coefficients of the characteristic equation @xmath672 as generators , it follows from that this map is given on the generic points in the base by @xmath673 where the sign in the last component is determined by the isomorphism @xmath674 .
for example , following the notation of remark [ detiso ] , if @xmath336 is the positively oriented local frame , then @xmath675 [ heuristic ] as explained at the beginning of section [ lie ] , the relation between spectral curves and thus the map can be understood heuristically from the relation between the eigenvalues of the @xmath678 and those of @xmath679 . by remark [ evalsadd ] , at least at smooth unramified points , the equations for the spectral curves @xmath19 and @xmath356 are given by @xmath680 from which the relations between @xmath681 and @xmath352 can be deduced . as described in
[ real so ] , the vector bundle of an @xmath307-higgs bundles may be expressed as @xmath682 , and thus they are labelled by two integer topological invariants , say @xmath683 , corresponding to the degrees of the line bundles @xmath325 and @xmath326 ( or , equivalently by classes in @xmath684 ) .
the moduli space @xmath685 is thus a disjoint union of ( possibly disconnected or empty ) components @xmath686 the moduli space @xmath687 is similarly a disjoint union of ( possibly disconnected or empty ) components @xmath688 for @xmath689 the degrees of the line bundles defining the @xmath341-higgs bundles . as seen in @xcite ,
the components are non - empty if and only if @xmath690 , inequalities which are known as the the milnor - wood bounds . where @xmath699 are line bundles defining @xmath304-higgs bundles ( see section [ iso22higgs ] ) .
the bounds on @xmath700 follow from the milnor - wood bound for @xmath67-higgs bundles .
the rest of the proposition follows from the observation that the preimage under @xmath71 for any higgs bundle of the form consists of all @xmath701-higgs bundles defined by @xmath702 with @xmath703 and @xmath704 . if @xmath705 then there are @xmath38 solutions for @xmath706 and @xmath707 . whilst the moduli space of @xmath46-higgs bundles has @xmath38 hitchin components , the one for @xmath412-higgs bundles has just one hitchin component . from the analysis of topological invariants , which are constant on connected components
, one has @xmath708 components , 4 coming from the 4 pairs of @xmath709 s characterizing @xmath448 bundles .
the isogeny @xmath71 between moduli spaces of higgs bundles takes the @xmath38 hitchin components to the one hitchin component , and the other 2 components to the two components ( possibly disconnected ) where the two @xmath709 s are the same.[hola ] as seen before , the map @xmath71 constructed in [ isosingular ]
is surjective onto some of the components of @xmath324 .
the components in the image correspond to those components in the representation variety @xmath710 for which the representations lift to @xmath341 .
we shall denote by @xmath711 the union of components of @xmath324 obtained through ( [ i*bundles ] ) and ( [ i*phi ] ) with @xmath712 equivalently , let @xmath713 be the union of components of @xmath710 which correspond to the components in @xmath714 .
[ cor1 ] the structure group of an @xmath68-higgs bundle lifts to @xmath341 if and only if the @xmath68-higgs bundle lies in @xmath711 .
equivalently , a reductive surface group representation into @xmath68 lifts to a representation into @xmath341 if and only if the representation lies @xmath713 . by realising @xmath307-higgs bundles in terms of rank 2-higgs bundles
, one can understand the monodromy action studied in @xcite for rank 4 higgs bundles in terms of monodromy of lower rank higgs bundles .
indeed , taking @xmath715 and @xmath318 as in eq .
one recovers the rank 4 monodromy as a product of actions coming from the rank 2-hitchin systems . to fully specify
the map induced by @xmath73 ( as in ) on @xmath659 one needs to specify the trivialization @xmath716 .
then , using the coefficients of the characteristic equation @xmath672 as generators for the rings of invariant polynomials , it follows from that the map on the generic points of the hitchin base is @xmath717 where the sign in the second component is determined by the isomorphism @xmath674 . for both n=2,3 the variety @xmath718 has two components . in the case @xmath667 , given @xmath61 the spectral curve defined by @xmath719 , the two components occur in the fibers over both points @xmath720 in the hitchin base . however , on each fiber only one of the components is in the image of the map induced by @xmath73 .
a similar phenomenon occurs for @xmath383 . as a consequence of its special orthogonal structure ,
the vector bundle in an @xmath46-higgs bundle is classified topologically by a second stiefel - whitney class .
the moduli space @xmath721 thus decomposes into components ( not necessarily connected ) labelled by this @xmath722-valued invariant .
the moduli space @xmath66 likewise has a decomposition into components labelled by a pair of @xmath722-valued invariants corresponding to the second stiefel - whitney classes of the two rank three bundles . using superscripts to indicate these characteristic classes ,
we can thus write @xmath723 it follows from proposition [ reali3 ] that the image of @xmath729 lies in the component @xmath730 , where @xmath731 and @xmath732 for some @xmath0 vector bundle @xmath55 with @xmath733 .
moreover , from ( * ? ? ?
* proposition 1.8 ) it follows that @xmath734 , so in particular @xmath735 .
however , any pair of @xmath736-bundles with the same second stieffel - whitney class arises in this way , i.e .. as @xmath737 where @xmath55 is an @xmath0 bundle . finally ,
if @xmath738 represents a point in the image of @xmath73 , then the @xmath38 preimages come from twisting @xmath55 by any point of order two in @xmath739 .
aparicio arroyo , m._the geometry of @xmath740-higgs bundles _ , ph.d .
thesis ( 2009 ) , universidad de salamanca .
baraglia d. , schaposnik l.p .
_ monodromy of rank 2 twisted hitchin systems and real character varieties _ , arxiv:1506.00372 ( 2015 ) .
beauville a. , m.s .
narasimhan and s. ramanan , _ spectral curves and the generalised theta divisor _
, j. reine angew . math .
398 , ( 1989 ) , 169 - 179 .
donagi r. , _ spectral covers _ , msri series ( 1995 ) , 28 , 65 - 86 .
donagi r. , _ the tetragonal construction _ , bulletin of ams ( 1981 ) , vol 4 no .
2 . gothen p. b. , _ the topology of higgs bundle moduli spaces _
thesis ( 1995 ) , university of warwick .
grove k. , ziller w. , _ lifting group actions and non - negative curvature _ , trans .
( 2011 ) , 28652890 .
helgason s. , _ differential geometry , lie groups and symmetric spaces _
ac . press ( 1978 ) , san diego .
hitchin n.j .
, _ the self - duality equations on a riemann surface _
lms , 55 ( 1987 ) , 3 .
hitchin n.j . , _ stable bundles and integrable systems _ , duke math .
j. , 54 1 .
( 1987 ) , 91 - 114 .
hitchin n.j . , _ lie groups and teichmller space _ , topology 31 , 3 , ( 1992 ) , 449 - 473 .
hitchin n.j . , _
higgs bundles and characteristic classes _ , ( 2013 ) , arxiv:1308.4603 .
hitchin n.j . , schaposnik l.p .
_ nonabelianization of higgs bundles _ , j. diff .
( 2014 ) manetti m. , _ iterated double covers and connected components of moduli spaces _ , topology 36 , ( 1996 ) , 745 - 764 .
mumford d. , _ theta characteristics of an algebraic curve _ , ann .
sci . cole norm .
( 4 ) * 4 * , ( 1971 ) , 181192 .
schaposnik l.p .
, _ monodromy of the @xmath743 hitchin fibration _ ,
, 24 2 , ( 2013 ) .
schaposnik l.p .
, _ spectral data for @xmath5-higgs bundles _ , d.phil .
thesis ( 2013 ) , university of oxford .
schaposnik l.p .
, _ spectral data for u(m , m)-higgs bundles_. to appear in _ int .
not_. schaposnik l.p .
, _ an introduction to spectral data for higgs bundles _ , ( 2014 ) , arxiv:1408.0333 . | we explore relations between higgs bundles that result from isogenies between low - dimensional lie groups , with special attention to the spectral data for the higgs bundles .
we focus on isogenies onto @xmath0 and @xmath1 and their split real forms . using fiber products of spectral curves ,
we obtain directly the desingularizations of the ( necessarily singular ) spectral curves associated to orthogonal higgs bundles . in the case of @xmath1
our construction can be interpreted as a new description of recillas trigonal construction .
exceptional isomorphisms between low rank lie algebras , and the corresponding isogenies between lie groups , have been a source of fascination since first catalogued by cartan a century ago ( @xcite , see also @xcite ) . in this paper
we explore the implications for higgs bundles of the isogenies between lie groups of rank 2 and 3 : @xmath2 as well as their restrictions to the split real forms of these complex semsimple groups .
higgs bundles over a compact riemann surface @xmath3 of genus @xmath4 were introduced in @xcite and in a more general setting in @xcite . for any lie group @xmath5 ,
a @xmath5-higgs bundle on @xmath3 is a pair @xmath6 where @xmath7 is a holomorphic principal bundle and @xmath8 ( the higgs field ) is a holomorphic section of an associated bundle twisted by @xmath9 , the canonical bundle of the surface @xmath3 .
if @xmath5 is a complex group then @xmath7 is a principal @xmath5-bundle , but if @xmath5 is a real form of a complex group then the structure group of @xmath7 is the complexification of a maximal compact subgroup of @xmath5 . in this paper
we consider only matrix groups , in particular @xmath10 or @xmath11 , and real forms of these groups . in these cases ( described in [ fibration ] ) the higgs bundles can be seen as holomorphic vector bundles with extra structure , where the precise nature of the extra structure is determined by the group @xmath5 , and the higgs fields are appropriately constrained sections of the endomorphism bundle twisted by @xmath9 .
the defining data for @xmath5-higgs bundle serves to construct @xmath5-local systems on @xmath3 .
indeed this is the crux of non - abelian hodge theory ( naht ) , whereby the moduli space of semistable @xmath5-higgs bundles on @xmath3 is identified with the moduli space of reductive representations of @xmath12 into @xmath5 .
the implications of any group homomorphism @xmath13 are clear for surface group representations , since composition with @xmath14 induces a map from representations into @xmath15 to representations into @xmath16 .
it follows from naht that there must be a corresponding induced map between the higgs bundle moduli spaces , but the transcendental nature of the naht correspondence means that the clarity of the induced map on representations does not transfer so easily to the induced map on higgs bundles .
our goal is to understand this map in the cases where the group homomorphism is given by the above isogenies .
in @xcite hitchin showed how the defining data of higgs bundles can be re - encoded into a so - called spectral data set consisting of a ramified covering @xmath17 of @xmath3 and a bundle on @xmath17 .
for the groups of interest in this paper , the spectral bundles are line bundles , and hence lie in jacobians of the spectral curve .
in fact , they must lie in prym varieties determined by the @xmath17 and @xmath3 .
these abelian varieties form the generic fibers in a fibration of the higgs bundle moduli space over a half - dimensional linear space . in terms of vector bundles ,
the effect of some low rank isogenies has been studied in @xcite . in this paper
we expand those result in the case of through the use of hodge star type operators . for both and we give a novel description of the spectral data correspondences , shedding new light on the maps between higgs bundles and their moduli spaces .
our key tool for understanding the induced maps on the spectral data for the higgs bundles is a fibre - product construction , at the heart of which is the following diagram : @xmath18&\cl = p_1^*l_1\otimes p_2^*l_2\ar[d]&l_2\ar[d]\\ s_1\ar[dr]_{\pi_1}&\ar[l]_{p_1}s_1\times_{\sigma}s_2\ar[d]^{\pi}\ar[r]^{p_2}&s_2\ar[dl]^{\pi_2}\\ & \sigma & } \ ] ] in our applications , the curves @xmath19 and line bundles @xmath20 come from spectral data for @xmath21 or @xmath22-higgs bundles , while the fiber - produc @xmath23 , and the line bundle @xmath24 yield the spectral data sets for the @xmath0 or @xmath1-higgs bundles .
this construction clearly has wider applicability than the cases inspired by the isogenies and .
one notable feature of the construction is that it generically yields smooth curves , even though curves defined by spectral data for @xmath11-higgs bundles are necessarily singular .
indeed , the curves provided by our construction are the normalizations of the singular @xmath11-spectral curves .
as seen in @xcite , the isogeny @xmath25 innduces a map on @xmath26-higgs bundles @xmath27 given by @xmath28 .
\end{aligned}\ ] ] where @xmath29 has orthogonal structure determined by the symplectic structures on @xmath30 . hence ,
in the case of the rank 2 isogeny @xmath25 , on the generic fibers of the hitchin fibration for the moduli space @xmath31 , our main result is the following ( see propositions [ i2spec ] , [ i2realhiggs ] , [ season2 ] ) : [ teorema2 ] let @xmath19 be the spectral curve of the @xmath26-higgs bundles @xmath27 , and @xmath32 the corresponding spectral line bundle .
then the spectral data for the the image @xmath33 is given by @xmath34 where * @xmath35 is a smooth ramified fourfold cover , and * @xmath36 where @xmath37 are the projection maps .
the map is a @xmath38 fold - coverings onto its images .
restricted to @xmath39-higgs bundles , the image lies only in components of @xmath40 in which the higgs bundles satisfy a topological constraint . in the case of the rank 3 isogeny , as seen in @xcite , the map @xmath41 induces a morphism on @xmath22-higgs bundles @xmath42 given by @xmath43 . \ ] ] where @xmath44 has orthogonal structure determined by the isomorphism @xmath45 .
this case differs from the previous one in two notable ways .
the first is seen in the restriction of to @xmath46-higgs bundles , where the map reflects the fact that @xmath47 is not a simple lie group .
this leads to a decomposition of the bundle @xmath44 that is analogous to the decomposition of 2-forms on a riemannian 4-manifold into self - dual and anti - self - dual forms .
the other new wrinkle is in the induced map on spectral data where the fiber product in diagram is now taken on two copies of the same covering , say @xmath17 .
the resulting curve can never be smooth since it has a diagonally embedded copy of @xmath17 which intersects other components at the ramification points for the covering @xmath48 .
it turns out though that this component plays no part in the map induced by @xmath49 .
we thus modify our construction by removing the diagonally embedded component in the fiber product of an @xmath22-spectral data set with itself ( see section [ isosmooth ] , proposition [ reali3 ] , [ rank3main ] , [ propinv1 ] for details ) : [ teorema1 ] restricted to the higgs bundles for split real form @xmath50 , the induced map yields @xmath51 . \nonumber\end{aligned}\ ] ] where @xmath52 are the @xmath53-eigenbundles for an involution @xmath54 determined by an orthogonal structure on @xmath55 , with canonically determined orthogonal structures , and with @xmath8 and @xmath56 related as in eq . .
on the generic fibers of the hitchin fibration for the moduli space @xmath57 , given the spectral data @xmath58 corresponding to @xmath42 , the spectral data corresponding to the @xmath1-higgs bundle @xmath59 is @xmath60 , where * @xmath61 is the symmetrization of the non - diagonal component in the fibre product @xmath62 ; * @xmath63 is a canonical twist of the line bundle generated by local sections of @xmath64 that are anti - invariant with respect to the symmetry of @xmath62 . the map is a @xmath38 fold - coverings onto its images .
restricted to points representing @xmath65-higgs bundles , the image lies only in components of @xmath66 in which the higgs bundles satisfy a topological constraint . while the two cases covered by and are by no means the only interesting ones , they have some unique features and serve to illustrate phenomena that we expect to apply in greater generality .
the group @xmath67 has a distinguished place in any discussion of surface group representations because of its relation to hyperbolic structures and teichmuller space , while the group @xmath68 is the isomtery group of the anti - de sitter space @xmath69 . moreover ,
both groups are split real forms of complex semisimple groups ( viz . @xmath21 and @xmath0 respectively ) and also
groups of hermitian type .
the only other such group is @xmath70 . for both classes of real lie group ,
the representation variety or , equivalently , the moduli space of higgs bundles has a distinguished set of components . in the case of the split real forms ,
the distinguished components are called `` higher teichmuller components '' because they generalize the copies of teichmuller space which occur in the case of @xmath67 . for the groups of hermitian type , the @xmath5-higgs bundles ( or the surface group representations ) carry a discrete invariant known as a toledo invariant .
this invariant satisfies a so - called milnor - wood bound and the distinguished components are those in which the invariant has maximal value .
in the two cases considered in this article , the induced maps on the spectral curves yield a pairing between two different ramified coverings of a common base curve , together with isogenies between the associated prym varieties .
the spectral curves in each pair are , moreover , determined by different representations of the same group .
this kind of situation , with pairings between coverings of a given curve and isogenies between the associated prym varieties , has been considered in the context of integrable systems and also by donagi ( see e.g. @xcite ) . as noted in (
* ? ? ?
* example 3 ) the correspondence we see for the pair @xmath22 and @xmath1 , namely between a four - fold covering and a six - fold covering with a fixed - point - free involution , is essentially the correspondence described by recillas in his trigonal construction @xcite and generalized by donagi in @xcite .
the novelty in our version of the correspondence and resulting relation between prym varieties lies in the use of fiber products to get explicit descriptions of the maps .
the first part of this paper covers background material on higgs bundles ( section [ fibration ] ) , spectral curves ( section [ spectral_higgs ] ) and the isogenies ( section [ lie ] ) . in section [ isosingular ]
we describe the maps induced by the isogeny @xmath71 on both the complex group @xmath72 and its split real form , and in section [ isosmooth ] we do the same for @xmath73 .
we include a discussion of the relation between our construction on spectral data and the trigonal construction of recillas , and show how the map we obtain between prym varieties can be interpreted in terms of a correspondence between curves .
we conclude , in section [ maps ] , with a discussion of maps between moduli spaces .
we thank david baraglia , ron donagi , tamas hausel , nigel hitchin , sheldon katz , herbert lange , tom nevins , tony pantev , mihnea popa , s. ramanan , hal schenck , and michael thaddeus , for helpful and enlightening discussions . both authors acknowledge support from u.s . national science foundation grants
dms 1107452 , 1107263 , 1107367 rnms : geometric structures and representation varieties ( the gear network ) . " |
the theory of graphons has provided a powerful tool for sampling and studying convergence properties for sequences of dense graphs .
graphons characterize limiting properties of dense graph sequences , such as properties arising in combinatorial optimization and statistical physics .
furthermore , sequences of dense graphs sampled from a ( possibly random ) graphon are characterized by a natural notion of exchangeability via the aldous - hoover theorem .
this paper presents an analogous theory for sparse graphs . in the past few years
, graphons have been used as non - parametric extensions of stochastic block models , to model and learn large networks .
there have been several rigorous papers on the subject of consistent estimation using graphons ( see , for example , @xcite and references therein ) , and graphons have also been used to estimate real - world networks , such as facebook and linkedin @xcite .
this makes it especially useful to have graphon models for sparse networks with unbounded degrees , which are the appropriate description of many large real - world networks . in the classical theory of graphons
as studied in , for example , @xcite , a graphon is a symmetric @xmath10$]-valued function defined on a probability space . in our generalized theory we let the underlying measure space of the graphon be a @xmath1-finite measure space ; i.e. , we allow the space to have infinite total measure . more precisely , given a @xmath1-finite measure space @xmath11 we define a graphon to be a pair @xmath12 , where @xmath13 is a symmetric integrable function , with the special case when @xmath14 is @xmath10$]-valued being most relevant for the random graphs studied in the current paper .
we present a random graph model associated with these generalized graphons which has a number of properties making it appropriate for modelling sparse networks , and we present a new theory for convergence of graphs in which our generalized graphons arise naturally as limits of sparse graphs . given a @xmath10$]-valued graphon @xmath12 with @xmath11 a @xmath1-finite measure space , we will now define a random process which generalizes the classical notion of @xmath15-random graphs , introduced in the statistics literature @xcite under the name latent position graphs , in the context of graph limits @xcite as @xmath15-random graphs , and in the context of extensions of the classical random graph theory @xcite as inhomogeneous random graphs . recall that in the classical setting where @xmath15 is defined on a probability space , @xmath15-random graphs are generated by first choosing @xmath16 points @xmath17 i.i.d . from the probability distribution @xmath18 over the feature space @xmath6 , and then connecting the vertices @xmath19 and @xmath20 with probability @xmath21 .
here , inspired by caron and fox @xcite , we generalize this to arbitrary @xmath1-finite measure spaces by first considering a poisson point process @xmath22 with intensity @xmath7 on @xmath6 for any fixed @xmath23 , and then connecting two points @xmath24 in @xmath22 with probability @xmath21 .
as explained in the next paragraph , this leads to a family of graphs @xmath25 such that the graphs @xmath26 have almost surely at most countably infinitely many vertices and ( assuming appropriate integrability conditions on @xmath14 , e.g. , @xmath27 ) a finite number of edges . removing all isolated vertices from @xmath26
, we obtain a family of graphs @xmath28 that are almost surely finite .
we refer to the families @xmath29 and @xmath30 as _ graphon processes _ ; when it is necessary to distinguish the two , we call them graphon processes with or without isolated vertices , respectively . from a graphon @xmath12 , where @xmath11 is a @xmath1-finite measure space .
the two coordinate axes on the middle figure represent our feature space @xmath6 , where the red ( resp .
blue ) dots on the axes represent vertices born during @xmath31 $ ] ( resp . @xmath32 $ ] ) for @xmath33 , and the red ( resp .
blue ) dots in the interior of the first quadrant represent edges in @xmath34 for @xmath35 ( resp .
@xmath36 ) . the graph @xmath5 is an induced subgraph of a graph @xmath26 with infinitely many vertices in the case @xmath37 , such that @xmath5 is obtained from @xmath26 by removing isolated vertices . at time
@xmath38 the marginal law of the features of @xmath39 is the points of a poisson point process on @xmath6 with intensity @xmath7 .
two distinct vertices with features @xmath40 and @xmath41 , respectively , are connected to each other by an undirected edge with probability @xmath42 .
the coordinate axes on the right figure represent time @xmath43 .
we get the graph @xmath5 by considering the edges restricted to @xmath44 ^ 2 $ ] .
note that the coordinate axes in the right figure and the graphs @xmath26 in the left figure are slightly inaccurate if we assume @xmath37 , since in this case there are infinitely many isolated vertices in @xmath26 for each @xmath23 .
we have chosen to label the vertices by the order in which they appear in @xmath5 , where ties are resolved by considering the time the vertices were born , i.e. , by considering the time they appeared in @xmath26 . ] to interpret the graphon process @xmath30 as a family of growing graphs we will need to couple the graphs @xmath5 for different times @xmath38 . to this end , we consider a poisson point process @xmath45 on @xmath46 ( @xmath47 being equipped with the borel @xmath1-algebra and lebesgue measure ) .
each point @xmath48 of @xmath45 corresponds to a vertex of an infinite graph @xmath49 , where the coordinate @xmath50 is interpreted as the time the vertex is born and the coordinate @xmath40 describes a feature of the vertex .
two distinct vertices @xmath48 and @xmath51 are connected by an undirected edge with probability @xmath42 , independently for each possible pair of distinct vertices .
for each fixed time @xmath38 define a graph @xmath5 by considering the induced subgraph of @xmath49 corresponding to vertices which are born at time @xmath50 or earlier , where we do not include vertices which would be isolated in @xmath5 .
see figure [ fig1 ] for an illustration .
the family of growing graphs @xmath30 just described includes classical dense @xmath15-random graphs ( up to isolated vertices ) and the sparse graphs studied in @xcite as special cases , and is ( except for minor technical differences ) identical to the family of random graphs studied in @xcite , a paper which was written in parallel with our paper ; see our remark at the end of this introduction . the graphon process @xmath25 satisfies a natural notion of exchangeability .
roughly speaking , in our setting this means that the features of newly born vertices are homogeneous in time .
more precisely , it can be defined as joint exchangeability of a random measure in @xmath52 , where the two coordinates correspond to time , and each edge of the graph corresponds to a point mass .
we will prove that graphon processes as defined above , with @xmath14 integrable and possibly random , are characterized by exchangeability of the random measure in @xmath52 along with a certain regularity condition we call _ uniform regularity of tails_. see proposition [ prop5 ] .
this result is an analogue of the aldous - hoover theorem @xcite ( which characterizes @xmath15-random graphs over probability spaces as graphs which are invariant in law under permutation of their vertices ) in the setting of possibly sparse graphs satisfying the aforementioned regularity condition .
the graphon processes defined above also have a number of other properties making them particularly natural to model sparse graphs or networks .
they are suitable for modelling networks which grow over time since no additional rescaling parameters ( like the explicitly given density dependence on the number of vertices specified in @xcite ) are necessary ; all information about the random graph model is encoded by the graphon alone .
the graphs are _ projective _ in the sense that if @xmath53 the graph @xmath54 is an induced subgraph of @xmath5 .
finally , a closely related family of weighted graphs is proven in @xcite to have power law degree distribution for certain @xmath15 , and our graphon processes are expected to behave similarly .
the graphon processes studied in this paper have a different qualitative behavior than the sparse @xmath15-random graphs studied in @xcite ( see figure [ fig3 ] ) , with the only overlap of the two theories occurring when the graphs are dense .
if the sparsity of the graphs is caused by the degrees of the vertices being scaled down approximately uniformly over time , then the model in @xcite is most natural .
if the sparsity is caused by later vertices typically having lower connectivity probabilities than earlier vertices , then the model presented in this paper is most natural .
the sampling method we will use in the forthcoming paper @xcite generalizes both of these methods .
( 0,0)(5.00000,0 ) ; ( 5.00000,0)(5.00000,5.00000 ) ; ( 5.00000,5.00000)(0,5.00000 ) ; ( 0,5.00000)(0,0 ) ; ( 0,0.100000)(0,0.200000)(0.100000,0.200000)(0.100000,0.100000)cycle ; ( 0,0.400000)(0,0.500000)(0.100000,0.500000)(0.100000,0.400000)cycle ; ( 0,0.500000)(0,0.600000)(0.100000,0.600000)(0.100000,0.500000)cycle ; ( 0,0.900000)(0,1.00000)(0.100000,1.00000)(0.100000,0.900000)cycle ; ( 0,1.00000)(0,1.10000)(0.100000,1.10000)(0.100000,1.00000)cycle ; ( 0,1.50000)(0,1.60000)(0.100000,1.60000)(0.100000,1.50000)cycle ; ( 0,2.00000)(0,2.10000)(0.100000,2.10000)(0.100000,2.00000)cycle ; ( 0,2.10000)(0,2.20000)(0.100000,2.20000)(0.100000,2.10000)cycle ; ( 0,2.20000)(0,2.30000)(0.100000,2.30000)(0.100000,2.20000)cycle ; ( 0,2.50000)(0,2.60000)(0.100000,2.60000)(0.100000,2.50000)cycle ; ( 0,2.60000)(0,2.70000)(0.100000,2.70000)(0.100000,2.60000)cycle ; ( 0,3.20000)(0,3.30000)(0.100000,3.30000)(0.100000,3.20000)cycle ; ( 0,3.50000)(0,3.60000)(0.100000,3.60000)(0.100000,3.50000)cycle ; ( 0,3.70000)(0,3.80000)(0.100000,3.80000)(0.100000,3.70000)cycle ; ( 0,3.80000)(0,3.90000)(0.100000,3.90000)(0.100000,3.80000)cycle ; ( 0,4.00000)(0,4.10000)(0.100000,4.10000)(0.100000,4.00000)cycle ; ( 0,4.10000)(0,4.20000)(0.100000,4.20000)(0.100000,4.10000)cycle ; ( 0,4.30000)(0,4.40000)(0.100000,4.40000)(0.100000,4.30000)cycle ; ( 0,4.70000)(0,4.80000)(0.100000,4.80000)(0.100000,4.70000)cycle ; ( 0,4.80000)(0,4.90000)(0.100000,4.90000)(0.100000,4.80000)cycle ; ( 0.100000,0)(0.100000,0.100000)(0.200000,0.100000)(0.200000,0)cycle ; ( 0.100000,0.200000)(0.100000,0.300000)(0.200000,0.300000)(0.200000,0.200000)cycle ; ( 0.100000,0.500000)(0.100000,0.600000)(0.200000,0.600000)(0.200000,0.500000)cycle ; ( 0.100000,0.700000)(0.100000,0.800000)(0.200000,0.800000)(0.200000,0.700000)cycle ; ( 0.100000,0.800000)(0.100000,0.900000)(0.200000,0.900000)(0.200000,0.800000)cycle ; ( 0.100000,0.900000)(0.100000,1.00000)(0.200000,1.00000)(0.200000,0.900000)cycle ; ( 0.100000,1.00000)(0.100000,1.10000)(0.200000,1.10000)(0.200000,1.00000)cycle ; ( 0.100000,1.20000)(0.100000,1.30000)(0.200000,1.30000)(0.200000,1.20000)cycle ; ( 0.100000,1.30000)(0.100000,1.40000)(0.200000,1.40000)(0.200000,1.30000)cycle ; ( 0.100000,1.80000)(0.100000,1.90000)(0.200000,1.90000)(0.200000,1.80000)cycle ; ( 0.100000,2.30000)(0.100000,2.40000)(0.200000,2.40000)(0.200000,2.30000)cycle ; ( 0.100000,2.50000)(0.100000,2.60000)(0.200000,2.60000)(0.200000,2.50000)cycle ; ( 0.100000,3.50000)(0.100000,3.60000)(0.200000,3.60000)(0.200000,3.50000)cycle ; ( 0.200000,0.100000)(0.200000,0.200000)(0.300000,0.200000)(0.300000,0.100000)cycle ; ( 0.200000,0.300000)(0.200000,0.400000)(0.300000,0.400000)(0.300000,0.300000)cycle ; ( 0.200000,0.500000)(0.200000,0.600000)(0.300000,0.600000)(0.300000,0.500000)cycle ; ( 0.200000,2.60000)(0.200000,2.70000)(0.300000,2.70000)(0.300000,2.60000)cycle ; ( 0.200000,3.40000)(0.200000,3.50000)(0.300000,3.50000)(0.300000,3.40000)cycle ; ( 0.200000,4.20000)(0.200000,4.30000)(0.300000,4.30000)(0.300000,4.20000)cycle ; ( 0.200000,4.70000)(0.200000,4.80000)(0.300000,4.80000)(0.300000,4.70000)cycle ; ( 0.200000,4.80000)(0.200000,4.90000)(0.300000,4.90000)(0.300000,4.80000)cycle ; ( 0.300000,0.200000)(0.300000,0.300000)(0.400000,0.300000)(0.400000,0.200000)cycle ; ( 0.300000,0.400000)(0.300000,0.500000)(0.400000,0.500000)(0.400000,0.400000)cycle ; ( 0.300000,0.800000)(0.300000,0.900000)(0.400000,0.900000)(0.400000,0.800000)cycle ; ( 0.300000,1.40000)(0.300000,1.50000)(0.400000,1.50000)(0.400000,1.40000)cycle ; ( 0.300000,2.00000)(0.300000,2.10000)(0.400000,2.10000)(0.400000,2.00000)cycle ; ( 0.300000,2.10000)(0.300000,2.20000)(0.400000,2.20000)(0.400000,2.10000)cycle ; ( 0.300000,2.30000)(0.300000,2.40000)(0.400000,2.40000)(0.400000,2.30000)cycle ; ( 0.300000,3.10000)(0.300000,3.20000)(0.400000,3.20000)(0.400000,3.10000)cycle ; ( 0.300000,3.60000)(0.300000,3.70000)(0.400000,3.70000)(0.400000,3.60000)cycle ; ( 0.400000,0)(0.400000,0.100000)(0.500000,0.100000)(0.500000,0)cycle ; ( 0.400000,0.300000)(0.400000,0.400000)(0.500000,0.400000)(0.500000,0.300000)cycle ; ( 0.400000,1.10000)(0.400000,1.20000)(0.500000,1.20000)(0.500000,1.10000)cycle ; ( 0.400000,1.40000)(0.400000,1.50000)(0.500000,1.50000)(0.500000,1.40000)cycle ; ( 0.400000,2.70000)(0.400000,2.80000)(0.500000,2.80000)(0.500000,2.70000)cycle ; ( 0.400000,3.50000)(0.400000,3.60000)(0.500000,3.60000)(0.500000,3.50000)cycle ; ( 0.500000,0)(0.500000,0.100000)(0.600000,0.100000)(0.600000,0)cycle ; ( 0.500000,0.100000)(0.500000,0.200000)(0.600000,0.200000)(0.600000,0.100000)cycle ; ( 0.500000,0.200000)(0.500000,0.300000)(0.600000,0.300000)(0.600000,0.200000)cycle ; ( 0.500000,0.700000)(0.500000,0.800000)(0.600000,0.800000)(0.600000,0.700000)cycle ; ( 0.500000,0.900000)(0.500000,1.00000)(0.600000,1.00000)(0.600000,0.900000)cycle ; ( 0.500000,1.20000)(0.500000,1.30000)(0.600000,1.30000)(0.600000,1.20000)cycle ; ( 0.500000,2.00000)(0.500000,2.10000)(0.600000,2.10000)(0.600000,2.00000)cycle ; ( 0.500000,2.10000)(0.500000,2.20000)(0.600000,2.20000)(0.600000,2.10000)cycle ; ( 0.500000,3.60000)(0.500000,3.70000)(0.600000,3.70000)(0.600000,3.60000)cycle ; ( 0.600000,0.700000)(0.600000,0.800000)(0.700000,0.800000)(0.700000,0.700000)cycle ; ( 0.600000,0.900000)(0.600000,1.00000)(0.700000,1.00000)(0.700000,0.900000)cycle ; ( 0.600000,1.10000)(0.600000,1.20000)(0.700000,1.20000)(0.700000,1.10000)cycle ; ( 0.600000,3.00000)(0.600000,3.10000)(0.700000,3.10000)(0.700000,3.00000)cycle ; ( 0.600000,3.40000)(0.600000,3.50000)(0.700000,3.50000)(0.700000,3.40000)cycle ; ( 0.600000,4.80000)(0.600000,4.90000)(0.700000,4.90000)(0.700000,4.80000)cycle ; ( 0.700000,0.100000)(0.700000,0.200000)(0.800000,0.200000)(0.800000,0.100000)cycle ; ( 0.700000,0.500000)(0.700000,0.600000)(0.800000,0.600000)(0.800000,0.500000)cycle ; ( 0.700000,0.600000)(0.700000,0.700000)(0.800000,0.700000)(0.800000,0.600000)cycle ; ( 0.700000,1.30000)(0.700000,1.40000)(0.800000,1.40000)(0.800000,1.30000)cycle ; ( 0.700000,1.60000)(0.700000,1.70000)(0.800000,1.70000)(0.800000,1.60000)cycle ; ( 0.700000,2.00000)(0.700000,2.10000)(0.800000,2.10000)(0.800000,2.00000)cycle ; ( 0.700000,3.10000)(0.700000,3.20000)(0.800000,3.20000)(0.800000,3.10000)cycle ; ( 0.700000,4.00000)(0.700000,4.10000)(0.800000,4.10000)(0.800000,4.00000)cycle ; ( 0.700000,4.80000)(0.700000,4.90000)(0.800000,4.90000)(0.800000,4.80000)cycle ; ( 0.800000,0.100000)(0.800000,0.200000)(0.900000,0.200000)(0.900000,0.100000)cycle ; ( 0.800000,0.300000)(0.800000,0.400000)(0.900000,0.400000)(0.900000,0.300000)cycle ; ( 0.800000,2.30000)(0.800000,2.40000)(0.900000,2.40000)(0.900000,2.30000)cycle ; ( 0.800000,2.70000)(0.800000,2.80000)(0.900000,2.80000)(0.900000,2.70000)cycle ; ( 0.800000,4.10000)(0.800000,4.20000)(0.900000,4.20000)(0.900000,4.10000)cycle ; ( 0.800000,4.80000)(0.800000,4.90000)(0.900000,4.90000)(0.900000,4.80000)cycle ; ( 0.900000,0)(0.900000,0.100000)(1.00000,0.100000)(1.00000,0)cycle ; ( 0.900000,0.100000)(0.900000,0.200000)(1.00000,0.200000)(1.00000,0.100000)cycle ; ( 0.900000,0.500000)(0.900000,0.600000)(1.00000,0.600000)(1.00000,0.500000)cycle ; ( 0.900000,0.600000)(0.900000,0.700000)(1.00000,0.700000)(1.00000,0.600000)cycle ; ( 0.900000,1.70000)(0.900000,1.80000)(1.00000,1.80000)(1.00000,1.70000)cycle ; ( 0.900000,3.60000)(0.900000,3.70000)(1.00000,3.70000)(1.00000,3.60000)cycle ; ( 0.900000,4.10000)(0.900000,4.20000)(1.00000,4.20000)(1.00000,4.10000)cycle ; ( 1.00000,0)(1.00000,0.100000)(1.10000,0.100000)(1.10000,0)cycle ; ( 1.00000,0.100000)(1.00000,0.200000)(1.10000,0.200000)(1.10000,0.100000)cycle ; ( 1.00000,2.10000)(1.00000,2.20000)(1.10000,2.20000)(1.10000,2.10000)cycle ; ( 1.00000,3.40000)(1.00000,3.50000)(1.10000,3.50000)(1.10000,3.40000)cycle ; ( 1.10000,0.400000)(1.10000,0.500000)(1.20000,0.500000)(1.20000,0.400000)cycle ; ( 1.10000,0.600000)(1.10000,0.700000)(1.20000,0.700000)(1.20000,0.600000)cycle ; ( 1.20000,0.100000)(1.20000,0.200000)(1.30000,0.200000)(1.30000,0.100000)cycle ; ( 1.20000,0.500000)(1.20000,0.600000)(1.30000,0.600000)(1.30000,0.500000)cycle ; ( 1.20000,1.30000)(1.20000,1.40000)(1.30000,1.40000)(1.30000,1.30000)cycle ; ( 1.20000,4.90000)(1.20000,5.00000)(1.30000,5.00000)(1.30000,4.90000)cycle ; ( 1.30000,0.100000)(1.30000,0.200000)(1.40000,0.200000)(1.40000,0.100000)cycle ; ( 1.30000,0.700000)(1.30000,0.800000)(1.40000,0.800000)(1.40000,0.700000)cycle ; ( 1.30000,1.20000)(1.30000,1.30000)(1.40000,1.30000)(1.40000,1.20000)cycle ; ( 1.30000,3.30000)(1.30000,3.40000)(1.40000,3.40000)(1.40000,3.30000)cycle ; ( 1.30000,4.90000)(1.30000,5.00000)(1.40000,5.00000)(1.40000,4.90000)cycle ; ( 1.40000,0.300000)(1.40000,0.400000)(1.50000,0.400000)(1.50000,0.300000)cycle ; ( 1.40000,0.400000)(1.40000,0.500000)(1.50000,0.500000)(1.50000,0.400000)cycle ; ( 1.40000,3.80000)(1.40000,3.90000)(1.50000,3.90000)(1.50000,3.80000)cycle ; ( 1.40000,4.40000)(1.40000,4.50000)(1.50000,4.50000)(1.50000,4.40000)cycle ; ( 1.50000,0)(1.50000,0.100000)(1.60000,0.100000)(1.60000,0)cycle ; ( 1.50000,1.80000)(1.50000,1.90000)(1.60000,1.90000)(1.60000,1.80000)cycle ; ( 1.60000,0.700000)(1.60000,0.800000)(1.70000,0.800000)(1.70000,0.700000)cycle ; ( 1.60000,2.60000)(1.60000,2.70000)(1.70000,2.70000)(1.70000,2.60000)cycle ; ( 1.60000,2.70000)(1.60000,2.80000)(1.70000,2.80000)(1.70000,2.70000)cycle ; ( 1.60000,3.10000)(1.60000,3.20000)(1.70000,3.20000)(1.70000,3.10000)cycle ; ( 1.70000,0.900000)(1.70000,1.00000)(1.80000,1.00000)(1.80000,0.900000)cycle ; ( 1.70000,2.00000)(1.70000,2.10000)(1.80000,2.10000)(1.80000,2.00000)cycle ; ( 1.70000,2.70000)(1.70000,2.80000)(1.80000,2.80000)(1.80000,2.70000)cycle ; ( 1.70000,3.30000)(1.70000,3.40000)(1.80000,3.40000)(1.80000,3.30000)cycle ; ( 1.80000,0.100000)(1.80000,0.200000)(1.90000,0.200000)(1.90000,0.100000)cycle ; ( 1.80000,1.50000)(1.80000,1.60000)(1.90000,1.60000)(1.90000,1.50000)cycle ; ( 1.80000,3.90000)(1.80000,4.00000)(1.90000,4.00000)(1.90000,3.90000)cycle ; ( 1.90000,2.00000)(1.90000,2.10000)(2.00000,2.10000)(2.00000,2.00000)cycle ; ( 1.90000,2.80000)(1.90000,2.90000)(2.00000,2.90000)(2.00000,2.80000)cycle ; ( 1.90000,2.90000)(1.90000,3.00000)(2.00000,3.00000)(2.00000,2.90000)cycle ; ( 2.00000,0)(2.00000,0.100000)(2.10000,0.100000)(2.10000,0)cycle ; ( 2.00000,0.300000)(2.00000,0.400000)(2.10000,0.400000)(2.10000,0.300000)cycle ; ( 2.00000,0.500000)(2.00000,0.600000)(2.10000,0.600000)(2.10000,0.500000)cycle ; ( 2.00000,0.700000)(2.00000,0.800000)(2.10000,0.800000)(2.10000,0.700000)cycle ; ( 2.00000,1.70000)(2.00000,1.80000)(2.10000,1.80000)(2.10000,1.70000)cycle ; ( 2.00000,1.90000)(2.00000,2.00000)(2.10000,2.00000)(2.10000,1.90000)cycle ; ( 2.00000,3.70000)(2.00000,3.80000)(2.10000,3.80000)(2.10000,3.70000)cycle ; ( 2.00000,4.10000)(2.00000,4.20000)(2.10000,4.20000)(2.10000,4.10000)cycle ; ( 2.10000,0)(2.10000,0.100000)(2.20000,0.100000)(2.20000,0)cycle ; ( 2.10000,0.300000)(2.10000,0.400000)(2.20000,0.400000)(2.20000,0.300000)cycle ; ( 2.10000,0.500000)(2.10000,0.600000)(2.20000,0.600000)(2.20000,0.500000)cycle ; ( 2.10000,1.00000)(2.10000,1.10000)(2.20000,1.10000)(2.20000,1.00000)cycle ; ( 2.10000,2.30000)(2.10000,2.40000)(2.20000,2.40000)(2.20000,2.30000)cycle ; ( 2.10000,2.50000)(2.10000,2.60000)(2.20000,2.60000)(2.20000,2.50000)cycle ; ( 2.10000,3.60000)(2.10000,3.70000)(2.20000,3.70000)(2.20000,3.60000)cycle ; ( 2.10000,4.30000)(2.10000,4.40000)(2.20000,4.40000)(2.20000,4.30000)cycle ; ( 2.10000,4.80000)(2.10000,4.90000)(2.20000,4.90000)(2.20000,4.80000)cycle ; ( 2.20000,0)(2.20000,0.100000)(2.30000,0.100000)(2.30000,0)cycle ; ( 2.30000,0.100000)(2.30000,0.200000)(2.40000,0.200000)(2.40000,0.100000)cycle ; ( 2.30000,0.300000)(2.30000,0.400000)(2.40000,0.400000)(2.40000,0.300000)cycle ; ( 2.30000,0.800000)(2.30000,0.900000)(2.40000,0.900000)(2.40000,0.800000)cycle ; ( 2.30000,2.10000)(2.30000,2.20000)(2.40000,2.20000)(2.40000,2.10000)cycle ; ( 2.30000,2.90000)(2.30000,3.00000)(2.40000,3.00000)(2.40000,2.90000)cycle ; ( 2.30000,4.60000)(2.30000,4.70000)(2.40000,4.70000)(2.40000,4.60000)cycle ; ( 2.40000,3.40000)(2.40000,3.50000)(2.50000,3.50000)(2.50000,3.40000)cycle ; ( 2.40000,4.10000)(2.40000,4.20000)(2.50000,4.20000)(2.50000,4.10000)cycle ; ( 2.50000,0)(2.50000,0.100000)(2.60000,0.100000)(2.60000,0)cycle ; ( 2.50000,0.100000)(2.50000,0.200000)(2.60000,0.200000)(2.60000,0.100000)cycle ; ( 2.50000,2.10000)(2.50000,2.20000)(2.60000,2.20000)(2.60000,2.10000)cycle ; ( 2.50000,3.70000)(2.50000,3.80000)(2.60000,3.80000)(2.60000,3.70000)cycle ; ( 2.60000,0)(2.60000,0.100000)(2.70000,0.100000)(2.70000,0)cycle ; ( 2.60000,0.200000)(2.60000,0.300000)(2.70000,0.300000)(2.70000,0.200000)cycle ; ( 2.60000,1.60000)(2.60000,1.70000)(2.70000,1.70000)(2.70000,1.60000)cycle ; ( 2.70000,0.400000)(2.70000,0.500000)(2.80000,0.500000)(2.80000,0.400000)cycle ; ( 2.70000,0.800000)(2.70000,0.900000)(2.80000,0.900000)(2.80000,0.800000)cycle ; ( 2.70000,1.60000)(2.70000,1.70000)(2.80000,1.70000)(2.80000,1.60000)cycle ; ( 2.70000,1.70000)(2.70000,1.80000)(2.80000,1.80000)(2.80000,1.70000)cycle ; ( 2.80000,1.90000)(2.80000,2.00000)(2.90000,2.00000)(2.90000,1.90000)cycle ; ( 2.80000,3.70000)(2.80000,3.80000)(2.90000,3.80000)(2.90000,3.70000)cycle ; ( 2.90000,1.90000)(2.90000,2.00000)(3.00000,2.00000)(3.00000,1.90000)cycle ; ( 2.90000,2.30000)(2.90000,2.40000)(3.00000,2.40000)(3.00000,2.30000)cycle ; ( 2.90000,4.30000)(2.90000,4.40000)(3.00000,4.40000)(3.00000,4.30000)cycle ; ( 3.00000,0.600000)(3.00000,0.700000)(3.10000,0.700000)(3.10000,0.600000)cycle ; ( 3.10000,0.300000)(3.10000,0.400000)(3.20000,0.400000)(3.20000,0.300000)cycle ; ( 3.10000,0.700000)(3.10000,0.800000)(3.20000,0.800000)(3.20000,0.700000)cycle ; ( 3.10000,1.60000)(3.10000,1.70000)(3.20000,1.70000)(3.20000,1.60000)cycle ; ( 3.20000,0)(3.20000,0.100000)(3.30000,0.100000)(3.30000,0)cycle ; ( 3.30000,1.30000)(3.30000,1.40000)(3.40000,1.40000)(3.40000,1.30000)cycle ; ( 3.30000,1.70000)(3.30000,1.80000)(3.40000,1.80000)(3.40000,1.70000)cycle ; ( 3.30000,3.80000)(3.30000,3.90000)(3.40000,3.90000)(3.40000,3.80000)cycle ; ( 3.30000,4.50000)(3.30000,4.60000)(3.40000,4.60000)(3.40000,4.50000)cycle ; ( 3.40000,0.200000)(3.40000,0.300000)(3.50000,0.300000)(3.50000,0.200000)cycle ; ( 3.40000,0.600000)(3.40000,0.700000)(3.50000,0.700000)(3.50000,0.600000)cycle ; ( 3.40000,1.00000)(3.40000,1.10000)(3.50000,1.10000)(3.50000,1.00000)cycle ; ( 3.40000,2.40000)(3.40000,2.50000)(3.50000,2.50000)(3.50000,2.40000)cycle ; ( 3.50000,0)(3.50000,0.100000)(3.60000,0.100000)(3.60000,0)cycle ; ( 3.50000,0.100000)(3.50000,0.200000)(3.60000,0.200000)(3.60000,0.100000)cycle ; ( 3.50000,0.400000)(3.50000,0.500000)(3.60000,0.500000)(3.60000,0.400000)cycle ; ( 3.50000,3.70000)(3.50000,3.80000)(3.60000,3.80000)(3.60000,3.70000)cycle ; ( 3.50000,4.10000)(3.50000,4.20000)(3.60000,4.20000)(3.60000,4.10000)cycle ; ( 3.60000,0.300000)(3.60000,0.400000)(3.70000,0.400000)(3.70000,0.300000)cycle ; ( 3.60000,0.500000)(3.60000,0.600000)(3.70000,0.600000)(3.70000,0.500000)cycle ; ( 3.60000,0.900000)(3.60000,1.00000)(3.70000,1.00000)(3.70000,0.900000)cycle ; ( 3.60000,2.10000)(3.60000,2.20000)(3.70000,2.20000)(3.70000,2.10000)cycle ; ( 3.60000,3.90000)(3.60000,4.00000)(3.70000,4.00000)(3.70000,3.90000)cycle ; ( 3.70000,0)(3.70000,0.100000)(3.80000,0.100000)(3.80000,0)cycle ; ( 3.70000,2.00000)(3.70000,2.10000)(3.80000,2.10000)(3.80000,2.00000)cycle ; ( 3.70000,2.50000)(3.70000,2.60000)(3.80000,2.60000)(3.80000,2.50000)cycle ; ( 3.70000,2.80000)(3.70000,2.90000)(3.80000,2.90000)(3.80000,2.80000)cycle ; ( 3.70000,3.50000)(3.70000,3.60000)(3.80000,3.60000)(3.80000,3.50000)cycle ; ( 3.80000,0)(3.80000,0.100000)(3.90000,0.100000)(3.90000,0)cycle ; ( 3.80000,1.40000)(3.80000,1.50000)(3.90000,1.50000)(3.90000,1.40000)cycle ; ( 3.80000,3.30000)(3.80000,3.40000)(3.90000,3.40000)(3.90000,3.30000)cycle ; ( 3.90000,1.80000)(3.90000,1.90000)(4.00000,1.90000)(4.00000,1.80000)cycle ; ( 3.90000,3.60000)(3.90000,3.70000)(4.00000,3.70000)(4.00000,3.60000)cycle ; ( 3.90000,4.50000)(3.90000,4.60000)(4.00000,4.60000)(4.00000,4.50000)cycle ; ( 4.00000,0)(4.00000,0.100000)(4.10000,0.100000)(4.10000,0)cycle ; ( 4.00000,0.700000)(4.00000,0.800000)(4.10000,0.800000)(4.10000,0.700000)cycle ; ( 4.10000,0)(4.10000,0.100000)(4.20000,0.100000)(4.20000,0)cycle ; ( 4.10000,0.800000)(4.10000,0.900000)(4.20000,0.900000)(4.20000,0.800000)cycle ; ( 4.10000,0.900000)(4.10000,1.00000)(4.20000,1.00000)(4.20000,0.900000)cycle ; ( 4.10000,2.00000)(4.10000,2.10000)(4.20000,2.10000)(4.20000,2.00000)cycle ; ( 4.10000,2.40000)(4.10000,2.50000)(4.20000,2.50000)(4.20000,2.40000)cycle ; ( 4.10000,3.50000)(4.10000,3.60000)(4.20000,3.60000)(4.20000,3.50000)cycle ; ( 4.20000,0.200000)(4.20000,0.300000)(4.30000,0.300000)(4.30000,0.200000)cycle ; ( 4.30000,0)(4.30000,0.100000)(4.40000,0.100000)(4.40000,0)cycle ; ( 4.30000,2.10000)(4.30000,2.20000)(4.40000,2.20000)(4.40000,2.10000)cycle ; ( 4.30000,2.90000)(4.30000,3.00000)(4.40000,3.00000)(4.40000,2.90000)cycle ; ( 4.40000,1.40000)(4.40000,1.50000)(4.50000,1.50000)(4.50000,1.40000)cycle ; ( 4.50000,3.30000)(4.50000,3.40000)(4.60000,3.40000)(4.60000,3.30000)cycle ; ( 4.50000,3.90000)(4.50000,4.00000)(4.60000,4.00000)(4.60000,3.90000)cycle ; ( 4.60000,2.30000)(4.60000,2.40000)(4.70000,2.40000)(4.70000,2.30000)cycle ; ( 4.70000,0)(4.70000,0.100000)(4.80000,0.100000)(4.80000,0)cycle ; ( 4.70000,0.200000)(4.70000,0.300000)(4.80000,0.300000)(4.80000,0.200000)cycle ; ( 4.80000,0)(4.80000,0.100000)(4.90000,0.100000)(4.90000,0)cycle ; ( 4.80000,0.200000)(4.80000,0.300000)(4.90000,0.300000)(4.90000,0.200000)cycle ; ( 4.80000,0.600000)(4.80000,0.700000)(4.90000,0.700000)(4.90000,0.600000)cycle ; ( 4.80000,0.700000)(4.80000,0.800000)(4.90000,0.800000)(4.90000,0.700000)cycle ; ( 4.80000,0.800000)(4.80000,0.900000)(4.90000,0.900000)(4.90000,0.800000)cycle ; ( 4.80000,2.10000)(4.80000,2.20000)(4.90000,2.20000)(4.90000,2.10000)cycle ; ( 4.90000,1.20000)(4.90000,1.30000)(5.00000,1.30000)(5.00000,1.20000)cycle ; ( 4.90000,1.30000)(4.90000,1.40000)(5.00000,1.40000)(5.00000,1.30000)cycle ; ( 2.50000,5.50000 ) node rescaled graphon on probability space ; ( 7,0)(12.0000,0 ) ; ( 12.0000,0)(12.0000,5.00000 ) ; ( 12.0000,5.00000)(7,5.00000 ) ; ( 7,5.00000)(7,0 ) ; ( 7,0.188679)(7,0.283019)(7.09434,0.283019)(7.09434,0.188679)cycle ; ( 7,0.283019)(7,0.377358)(7.09434,0.377358)(7.09434,0.283019)cycle ; ( 7,0.377358)(7,0.471698)(7.09434,0.471698)(7.09434,0.377358)cycle ; ( 7,0.471698)(7,0.566038)(7.09434,0.566038)(7.09434,0.471698)cycle ; ( 7,0.566038)(7,0.660377)(7.09434,0.660377)(7.09434,0.566038)cycle ; ( 7,0.660377)(7,0.754717)(7.09434,0.754717)(7.09434,0.660377)cycle ; ( 7,0.754717)(7,0.849057)(7.09434,0.849057)(7.09434,0.754717)cycle ; ( 7,0.849057)(7,0.943396)(7.09434,0.943396)(7.09434,0.849057)cycle ; ( 7,0.943396)(7,1.03774)(7.09434,1.03774)(7.09434,0.943396)cycle ; ( 7,1.03774)(7,1.13208)(7.09434,1.13208)(7.09434,1.03774)cycle ; ( 7,1.13208)(7,1.22642)(7.09434,1.22642)(7.09434,1.13208)cycle ; ( 7,1.22642)(7,1.32075)(7.09434,1.32075)(7.09434,1.22642)cycle ; ( 7,1.32075)(7,1.41509)(7.09434,1.41509)(7.09434,1.32075)cycle ; ( 7,2.16981)(7,2.26415)(7.09434,2.26415)(7.09434,2.16981)cycle ; ( 7,2.64151)(7,2.73585)(7.09434,2.73585)(7.09434,2.64151)cycle ; ( 7,3.01887)(7,3.11321)(7.09434,3.11321)(7.09434,3.01887)cycle ; ( 7,4.43396)(7,4.52830)(7.09434,4.52830)(7.09434,4.43396)cycle ; ( 7,4.52830)(7,4.62264)(7.09434,4.62264)(7.09434,4.52830)cycle ; ( 7.09434,0.188679)(7.09434,0.283019)(7.18868,0.283019)(7.18868,0.188679)cycle ; ( 7.09434,0.283019)(7.09434,0.377358)(7.18868,0.377358)(7.18868,0.283019)cycle ; ( 7.09434,0.377358)(7.09434,0.471698)(7.18868,0.471698)(7.18868,0.377358)cycle ; ( 7.09434,0.660377)(7.09434,0.754717)(7.18868,0.754717)(7.18868,0.660377)cycle ; ( 7.09434,0.754717)(7.09434,0.849057)(7.18868,0.849057)(7.18868,0.754717)cycle ; ( 7.09434,0.849057)(7.09434,0.943396)(7.18868,0.943396)(7.18868,0.849057)cycle ; ( 7.09434,0.943396)(7.09434,1.03774)(7.18868,1.03774)(7.18868,0.943396)cycle ; ( 7.09434,1.13208)(7.09434,1.22642)(7.18868,1.22642)(7.18868,1.13208)cycle ; ( 7.09434,1.32075)(7.09434,1.41509)(7.18868,1.41509)(7.18868,1.32075)cycle ; ( 7.09434,1.41509)(7.09434,1.50943)(7.18868,1.50943)(7.18868,1.41509)cycle ; ( 7.09434,1.69811)(7.09434,1.79245)(7.18868,1.79245)(7.18868,1.69811)cycle ; ( 7.09434,1.88679)(7.09434,1.98113)(7.18868,1.98113)(7.18868,1.88679)cycle ; ( 7.09434,2.26415)(7.09434,2.35849)(7.18868,2.35849)(7.18868,2.26415)cycle ; ( 7.09434,2.92453)(7.09434,3.01887)(7.18868,3.01887)(7.18868,2.92453)cycle ; ( 7.09434,4.15094)(7.09434,4.24528)(7.18868,4.24528)(7.18868,4.15094)cycle ; ( 7.09434,4.33962)(7.09434,4.43396)(7.18868,4.43396)(7.18868,4.33962)cycle ; ( 7.09434,4.62264)(7.09434,4.71698)(7.18868,4.71698)(7.18868,4.62264)cycle ; ( 7.09434,4.71698)(7.09434,4.81132)(7.18868,4.81132)(7.18868,4.71698)cycle ; ( 7.18868,0)(7.18868,0.0943396)(7.28302,0.0943396)(7.28302,0)cycle ; ( 7.18868,0.0943396)(7.18868,0.188679)(7.28302,0.188679)(7.28302,0.0943396)cycle ; ( 7.18868,0.283019)(7.18868,0.377358)(7.28302,0.377358)(7.28302,0.283019)cycle ; ( 7.18868,0.471698)(7.18868,0.566038)(7.28302,0.566038)(7.28302,0.471698)cycle ; ( 7.18868,0.566038)(7.18868,0.660377)(7.28302,0.660377)(7.28302,0.566038)cycle ; ( 7.18868,0.660377)(7.18868,0.754717)(7.28302,0.754717)(7.28302,0.660377)cycle ; ( 7.18868,0.943396)(7.18868,1.03774)(7.28302,1.03774)(7.28302,0.943396)cycle ; ( 7.18868,1.03774)(7.18868,1.13208)(7.28302,1.13208)(7.28302,1.03774)cycle ; ( 7.18868,1.32075)(7.18868,1.41509)(7.28302,1.41509)(7.28302,1.32075)cycle ; ( 7.18868,1.41509)(7.18868,1.50943)(7.28302,1.50943)(7.28302,1.41509)cycle ; ( 7.18868,1.60377)(7.18868,1.69811)(7.28302,1.69811)(7.28302,1.60377)cycle ; ( 7.18868,1.69811)(7.18868,1.79245)(7.28302,1.79245)(7.28302,1.69811)cycle ; ( 7.18868,2.07547)(7.18868,2.16981)(7.28302,2.16981)(7.28302,2.07547)cycle ; ( 7.18868,2.54717)(7.18868,2.64151)(7.28302,2.64151)(7.28302,2.54717)cycle ; ( 7.18868,2.73585)(7.18868,2.83019)(7.28302,2.83019)(7.28302,2.73585)cycle ; ( 7.18868,3.77358)(7.18868,3.86792)(7.28302,3.86792)(7.28302,3.77358)cycle ; ( 7.18868,3.96226)(7.18868,4.05660)(7.28302,4.05660)(7.28302,3.96226)cycle ; ( 7.18868,4.05660)(7.18868,4.15094)(7.28302,4.15094)(7.28302,4.05660)cycle ; ( 7.28302,0)(7.28302,0.0943396)(7.37736,0.0943396)(7.37736,0)cycle ; ( 7.28302,0.0943396)(7.28302,0.188679)(7.37736,0.188679)(7.37736,0.0943396)cycle ; ( 7.28302,0.188679)(7.28302,0.283019)(7.37736,0.283019)(7.37736,0.188679)cycle ; ( 7.28302,0.377358)(7.28302,0.471698)(7.37736,0.471698)(7.37736,0.377358)cycle ; ( 7.28302,0.471698)(7.28302,0.566038)(7.37736,0.566038)(7.37736,0.471698)cycle ; ( 7.28302,0.566038)(7.28302,0.660377)(7.37736,0.660377)(7.37736,0.566038)cycle ; ( 7.28302,0.660377)(7.28302,0.754717)(7.37736,0.754717)(7.37736,0.660377)cycle ; ( 7.28302,0.754717)(7.28302,0.849057)(7.37736,0.849057)(7.37736,0.754717)cycle ; ( 7.28302,0.849057)(7.28302,0.943396)(7.37736,0.943396)(7.37736,0.849057)cycle ; ( 7.28302,0.943396)(7.28302,1.03774)(7.37736,1.03774)(7.37736,0.943396)cycle ; ( 7.28302,1.03774)(7.28302,1.13208)(7.37736,1.13208)(7.37736,1.03774)cycle ; ( 7.28302,1.22642)(7.28302,1.32075)(7.37736,1.32075)(7.37736,1.22642)cycle ; ( 7.28302,1.32075)(7.28302,1.41509)(7.37736,1.41509)(7.37736,1.32075)cycle ; ( 7.28302,2.07547)(7.28302,2.16981)(7.37736,2.16981)(7.37736,2.07547)cycle ; ( 7.28302,2.16981)(7.28302,2.26415)(7.37736,2.26415)(7.37736,2.16981)cycle ; ( 7.28302,2.54717)(7.28302,2.64151)(7.37736,2.64151)(7.37736,2.54717)cycle ; ( 7.28302,3.11321)(7.28302,3.20755)(7.37736,3.20755)(7.37736,3.11321)cycle ; ( 7.28302,3.39623)(7.28302,3.49057)(7.37736,3.49057)(7.37736,3.39623)cycle ; ( 7.28302,4.24528)(7.28302,4.33962)(7.37736,4.33962)(7.37736,4.24528)cycle ; ( 7.28302,4.90566)(7.28302,5.00000)(7.37736,5.00000)(7.37736,4.90566)cycle ; ( 7.37736,0)(7.37736,0.0943396)(7.47170,0.0943396)(7.47170,0)cycle ; ( 7.37736,0.0943396)(7.37736,0.188679)(7.47170,0.188679)(7.47170,0.0943396)cycle ; ( 7.37736,0.283019)(7.37736,0.377358)(7.47170,0.377358)(7.47170,0.283019)cycle ; ( 7.37736,0.471698)(7.37736,0.566038)(7.47170,0.566038)(7.47170,0.471698)cycle ; ( 7.37736,0.566038)(7.37736,0.660377)(7.47170,0.660377)(7.47170,0.566038)cycle ; ( 7.37736,0.660377)(7.37736,0.754717)(7.47170,0.754717)(7.47170,0.660377)cycle ; ( 7.37736,0.754717)(7.37736,0.849057)(7.47170,0.849057)(7.47170,0.754717)cycle ; ( 7.37736,0.849057)(7.37736,0.943396)(7.47170,0.943396)(7.47170,0.849057)cycle ; ( 7.37736,0.943396)(7.37736,1.03774)(7.47170,1.03774)(7.47170,0.943396)cycle ; ( 7.37736,1.32075)(7.37736,1.41509)(7.47170,1.41509)(7.47170,1.32075)cycle ; ( 7.37736,1.41509)(7.37736,1.50943)(7.47170,1.50943)(7.47170,1.41509)cycle ; ( 7.37736,1.50943)(7.37736,1.60377)(7.47170,1.60377)(7.47170,1.50943)cycle ; ( 7.37736,1.60377)(7.37736,1.69811)(7.47170,1.69811)(7.47170,1.60377)cycle ; ( 7.37736,1.88679)(7.37736,1.98113)(7.47170,1.98113)(7.47170,1.88679)cycle ; ( 7.37736,1.98113)(7.37736,2.07547)(7.47170,2.07547)(7.47170,1.98113)cycle ; ( 7.37736,2.07547)(7.37736,2.16981)(7.47170,2.16981)(7.47170,2.07547)cycle ; ( 7.37736,2.45283)(7.37736,2.54717)(7.47170,2.54717)(7.47170,2.45283)cycle ; ( 7.37736,2.83019)(7.37736,2.92453)(7.47170,2.92453)(7.47170,2.83019)cycle ; ( 7.37736,3.30189)(7.37736,3.39623)(7.47170,3.39623)(7.47170,3.30189)cycle ; ( 7.37736,4.81132)(7.37736,4.90566)(7.47170,4.90566)(7.47170,4.81132)cycle ; ( 7.47170,0)(7.47170,0.0943396)(7.56604,0.0943396)(7.56604,0)cycle ; ( 7.47170,0.188679)(7.47170,0.283019)(7.56604,0.283019)(7.56604,0.188679)cycle ; ( 7.47170,0.283019)(7.47170,0.377358)(7.56604,0.377358)(7.56604,0.283019)cycle ; ( 7.47170,0.377358)(7.47170,0.471698)(7.56604,0.471698)(7.56604,0.377358)cycle ; ( 7.47170,0.660377)(7.47170,0.754717)(7.56604,0.754717)(7.56604,0.660377)cycle ; ( 7.47170,3.01887)(7.47170,3.11321)(7.56604,3.11321)(7.56604,3.01887)cycle ; ( 7.56604,0)(7.56604,0.0943396)(7.66038,0.0943396)(7.66038,0)cycle ; ( 7.56604,0.188679)(7.56604,0.283019)(7.66038,0.283019)(7.66038,0.188679)cycle ; ( 7.56604,0.283019)(7.56604,0.377358)(7.66038,0.377358)(7.66038,0.283019)cycle ; ( 7.56604,0.377358)(7.56604,0.471698)(7.66038,0.471698)(7.66038,0.377358)cycle ; ( 7.56604,0.849057)(7.56604,0.943396)(7.66038,0.943396)(7.66038,0.849057)cycle ; ( 7.56604,1.79245)(7.56604,1.88679)(7.66038,1.88679)(7.66038,1.79245)cycle ; ( 7.56604,3.11321)(7.56604,3.20755)(7.66038,3.20755)(7.66038,3.11321)cycle ; ( 7.56604,3.20755)(7.56604,3.30189)(7.66038,3.30189)(7.66038,3.20755)cycle ; ( 7.56604,3.86792)(7.56604,3.96226)(7.66038,3.96226)(7.66038,3.86792)cycle ; ( 7.66038,0)(7.66038,0.0943396)(7.75472,0.0943396)(7.75472,0)cycle ; ( 7.66038,0.0943396)(7.66038,0.188679)(7.75472,0.188679)(7.75472,0.0943396)cycle ; ( 7.66038,0.188679)(7.66038,0.283019)(7.75472,0.283019)(7.75472,0.188679)cycle ; ( 7.66038,0.283019)(7.66038,0.377358)(7.75472,0.377358)(7.75472,0.283019)cycle ; ( 7.66038,0.377358)(7.66038,0.471698)(7.75472,0.471698)(7.75472,0.377358)cycle ; ( 7.66038,0.471698)(7.66038,0.566038)(7.75472,0.566038)(7.75472,0.471698)cycle ; ( 7.66038,1.69811)(7.66038,1.79245)(7.75472,1.79245)(7.75472,1.69811)cycle ; ( 7.75472,0)(7.75472,0.0943396)(7.84906,0.0943396)(7.84906,0)cycle ; ( 7.75472,0.0943396)(7.75472,0.188679)(7.84906,0.188679)(7.84906,0.0943396)cycle ; ( 7.75472,0.283019)(7.75472,0.377358)(7.84906,0.377358)(7.84906,0.283019)cycle ; ( 7.75472,0.377358)(7.75472,0.471698)(7.84906,0.471698)(7.84906,0.377358)cycle ; ( 7.75472,1.88679)(7.75472,1.98113)(7.84906,1.98113)(7.84906,1.88679)cycle ; ( 7.75472,2.45283)(7.75472,2.54717)(7.84906,2.54717)(7.84906,2.45283)cycle ; ( 7.75472,2.54717)(7.75472,2.64151)(7.84906,2.64151)(7.84906,2.54717)cycle ; ( 7.84906,0)(7.84906,0.0943396)(7.94340,0.0943396)(7.94340,0)cycle ; ( 7.84906,0.0943396)(7.84906,0.188679)(7.94340,0.188679)(7.94340,0.0943396)cycle ; ( 7.84906,0.283019)(7.84906,0.377358)(7.94340,0.377358)(7.94340,0.283019)cycle ; ( 7.84906,0.377358)(7.84906,0.471698)(7.94340,0.471698)(7.94340,0.377358)cycle ; ( 7.84906,0.566038)(7.84906,0.660377)(7.94340,0.660377)(7.94340,0.566038)cycle ; ( 7.84906,1.03774)(7.84906,1.13208)(7.94340,1.13208)(7.94340,1.03774)cycle ; ( 7.94340,0)(7.94340,0.0943396)(8.03774,0.0943396)(8.03774,0)cycle ; ( 7.94340,0.0943396)(7.94340,0.188679)(8.03774,0.188679)(8.03774,0.0943396)cycle ; ( 7.94340,0.188679)(7.94340,0.283019)(8.03774,0.283019)(8.03774,0.188679)cycle ; ( 7.94340,0.283019)(7.94340,0.377358)(8.03774,0.377358)(8.03774,0.283019)cycle ; ( 7.94340,0.377358)(7.94340,0.471698)(8.03774,0.471698)(8.03774,0.377358)cycle ; ( 7.94340,3.67925)(7.94340,3.77358)(8.03774,3.77358)(8.03774,3.67925)cycle ; ( 8.03774,0)(8.03774,0.0943396)(8.13208,0.0943396)(8.13208,0)cycle ; ( 8.03774,0.188679)(8.03774,0.283019)(8.13208,0.283019)(8.13208,0.188679)cycle ; ( 8.03774,0.283019)(8.03774,0.377358)(8.13208,0.377358)(8.13208,0.283019)cycle ; ( 8.03774,0.849057)(8.03774,0.943396)(8.13208,0.943396)(8.13208,0.849057)cycle ; ( 8.13208,0)(8.13208,0.0943396)(8.22642,0.0943396)(8.22642,0)cycle ; ( 8.13208,0.0943396)(8.13208,0.188679)(8.22642,0.188679)(8.22642,0.0943396)cycle ; ( 8.13208,2.64151)(8.13208,2.73585)(8.22642,2.73585)(8.22642,2.64151)cycle ; ( 8.22642,0)(8.22642,0.0943396)(8.32075,0.0943396)(8.32075,0)cycle ; ( 8.22642,0.283019)(8.22642,0.377358)(8.32075,0.377358)(8.32075,0.283019)cycle ; ( 8.32075,0)(8.32075,0.0943396)(8.41509,0.0943396)(8.41509,0)cycle ; ( 8.32075,0.0943396)(8.32075,0.188679)(8.41509,0.188679)(8.41509,0.0943396)cycle ; ( 8.32075,0.188679)(8.32075,0.283019)(8.41509,0.283019)(8.41509,0.188679)cycle ; ( 8.32075,0.283019)(8.32075,0.377358)(8.41509,0.377358)(8.41509,0.283019)cycle ; ( 8.32075,0.377358)(8.32075,0.471698)(8.41509,0.471698)(8.41509,0.377358)cycle ; ( 8.32075,2.35849)(8.32075,2.45283)(8.41509,2.45283)(8.41509,2.35849)cycle ; ( 8.32075,2.45283)(8.32075,2.54717)(8.41509,2.54717)(8.41509,2.45283)cycle ; ( 8.41509,0.0943396)(8.41509,0.188679)(8.50943,0.188679)(8.50943,0.0943396)cycle ; ( 8.41509,0.188679)(8.41509,0.283019)(8.50943,0.283019)(8.50943,0.188679)cycle ; ( 8.41509,0.377358)(8.41509,0.471698)(8.50943,0.471698)(8.50943,0.377358)cycle ; ( 8.41509,3.58491)(8.41509,3.67925)(8.50943,3.67925)(8.50943,3.58491)cycle ; ( 8.50943,0.377358)(8.50943,0.471698)(8.60377,0.471698)(8.60377,0.377358)cycle ; ( 8.60377,0.188679)(8.60377,0.283019)(8.69811,0.283019)(8.69811,0.188679)cycle ; ( 8.60377,0.377358)(8.60377,0.471698)(8.69811,0.471698)(8.69811,0.377358)cycle ; ( 8.69811,0.0943396)(8.69811,0.188679)(8.79245,0.188679)(8.79245,0.0943396)cycle ; ( 8.69811,0.188679)(8.69811,0.283019)(8.79245,0.283019)(8.79245,0.188679)cycle ; ( 8.69811,0.660377)(8.69811,0.754717)(8.79245,0.754717)(8.79245,0.660377)cycle ; ( 8.69811,3.49057)(8.69811,3.58491)(8.79245,3.58491)(8.79245,3.49057)cycle ; ( 8.79245,0.566038)(8.79245,0.660377)(8.88679,0.660377)(8.88679,0.566038)cycle ; ( 8.88679,0.0943396)(8.88679,0.188679)(8.98113,0.188679)(8.98113,0.0943396)cycle ; ( 8.88679,0.377358)(8.88679,0.471698)(8.98113,0.471698)(8.98113,0.377358)cycle ; ( 8.88679,0.754717)(8.88679,0.849057)(8.98113,0.849057)(8.98113,0.754717)cycle ; ( 8.98113,0.377358)(8.98113,0.471698)(9.07547,0.471698)(9.07547,0.377358)cycle ; ( 8.98113,3.20755)(8.98113,3.30189)(9.07547,3.30189)(9.07547,3.20755)cycle ; ( 9.07547,0.188679)(9.07547,0.283019)(9.16981,0.283019)(9.16981,0.188679)cycle ; ( 9.07547,0.283019)(9.07547,0.377358)(9.16981,0.377358)(9.16981,0.283019)cycle ; ( 9.07547,0.377358)(9.07547,0.471698)(9.16981,0.471698)(9.16981,0.377358)cycle ; ( 9.16981,0)(9.16981,0.0943396)(9.26415,0.0943396)(9.26415,0)cycle ; ( 9.16981,0.283019)(9.16981,0.377358)(9.26415,0.377358)(9.26415,0.283019)cycle ; ( 9.26415,0.0943396)(9.26415,0.188679)(9.35849,0.188679)(9.35849,0.0943396)cycle ; ( 9.35849,1.32075)(9.35849,1.41509)(9.45283,1.41509)(9.45283,1.32075)cycle ; ( 9.45283,0.377358)(9.45283,0.471698)(9.54717,0.471698)(9.54717,0.377358)cycle ; ( 9.45283,0.754717)(9.45283,0.849057)(9.54717,0.849057)(9.54717,0.754717)cycle ; ( 9.45283,1.32075)(9.45283,1.41509)(9.54717,1.41509)(9.54717,1.32075)cycle ; ( 9.54717,0.188679)(9.54717,0.283019)(9.64151,0.283019)(9.64151,0.188679)cycle ; ( 9.54717,0.283019)(9.54717,0.377358)(9.64151,0.377358)(9.64151,0.283019)cycle ; ( 9.54717,0.754717)(9.54717,0.849057)(9.64151,0.849057)(9.64151,0.754717)cycle ; ( 9.64151,0)(9.64151,0.0943396)(9.73585,0.0943396)(9.73585,0)cycle ; ( 9.64151,1.13208)(9.64151,1.22642)(9.73585,1.22642)(9.73585,1.13208)cycle ; ( 9.73585,0.188679)(9.73585,0.283019)(9.83019,0.283019)(9.83019,0.188679)cycle ; ( 9.83019,0.377358)(9.83019,0.471698)(9.92453,0.471698)(9.92453,0.377358)cycle ; ( 9.92453,0.0943396)(9.92453,0.188679)(10.0189,0.188679)(10.0189,0.0943396)cycle ; ( 10.0189,0)(10.0189,0.0943396)(10.1132,0.0943396)(10.1132,0)cycle ; ( 10.0189,0.471698)(10.0189,0.566038)(10.1132,0.566038)(10.1132,0.471698)cycle ; ( 10.1132,0.283019)(10.1132,0.377358)(10.2075,0.377358)(10.2075,0.283019)cycle ; ( 10.1132,0.566038)(10.1132,0.660377)(10.2075,0.660377)(10.2075,0.566038)cycle ; ( 10.2075,0.566038)(10.2075,0.660377)(10.3019,0.660377)(10.3019,0.566038)cycle ; ( 10.2075,1.98113)(10.2075,2.07547)(10.3019,2.07547)(10.3019,1.98113)cycle ; ( 10.3019,0.377358)(10.3019,0.471698)(10.3962,0.471698)(10.3962,0.377358)cycle ; ( 10.3962,0.283019)(10.3962,0.377358)(10.4906,0.377358)(10.4906,0.283019)cycle ; ( 10.4906,1.69811)(10.4906,1.79245)(10.5849,1.79245)(10.5849,1.69811)cycle ; ( 10.5849,1.41509)(10.5849,1.50943)(10.6792,1.50943)(10.6792,1.41509)cycle ; ( 10.6792,0.943396)(10.6792,1.03774)(10.7736,1.03774)(10.7736,0.943396)cycle ; ( 10.7736,0.188679)(10.7736,0.283019)(10.8679,0.283019)(10.8679,0.188679)cycle ; ( 10.8679,0.566038)(10.8679,0.660377)(10.9623,0.660377)(10.9623,0.566038)cycle ; ( 10.9623,0.188679)(10.9623,0.283019)(11.0566,0.283019)(11.0566,0.188679)cycle ; ( 11.0566,0.188679)(11.0566,0.283019)(11.1509,0.283019)(11.1509,0.188679)cycle ; ( 11.1509,0.0943396)(11.1509,0.188679)(11.2453,0.188679)(11.2453,0.0943396)cycle ; ( 11.2453,0.283019)(11.2453,0.377358)(11.3396,0.377358)(11.3396,0.283019)cycle ; ( 11.3396,0.0943396)(11.3396,0.188679)(11.4340,0.188679)(11.4340,0.0943396)cycle ; ( 11.4340,0)(11.4340,0.0943396)(11.5283,0.0943396)(11.5283,0)cycle ; ( 11.5283,0)(11.5283,0.0943396)(11.6226,0.0943396)(11.6226,0)cycle ; ( 11.6226,0.0943396)(11.6226,0.188679)(11.7170,0.188679)(11.7170,0.0943396)cycle ; ( 11.7170,0.0943396)(11.7170,0.188679)(11.8113,0.188679)(11.8113,0.0943396)cycle ; ( 11.8113,0.377358)(11.8113,0.471698)(11.9057,0.471698)(11.9057,0.377358)cycle ; ( 11.9057,0.283019)(11.9057,0.377358)(12.0000,0.377358)(12.0000,0.283019)cycle ; ( 9.50000,5.50000 ) node graphon with non - compact support ; to compare different models , and to discuss notions of convergence , we introduce the following natural generalization of the cut metric for graphons on probability spaces to our setting . for two graphons @xmath55 and @xmath56 , this metric is easiest to define when the two graphons are defined over the same space .
however , for applications we want to compare graphons over different spaces , say two borel spaces @xmath57 and @xmath58 . assuming that both borel spaces have infinite total measure , the cut distance between @xmath59 and @xmath60
can then be defined as @xmath61 where we take the infimum over measure - preserving maps @xmath62 for @xmath63 , @xmath64 for @xmath65 , and the supremum is over measurable sets @xmath66 .
( see definition [ defn1 ] below for the definition of the cut distance for graphons over general spaces , including the case where one or both spaces have finite total mass . )
we call two graphons _ equivalent _ if they have cut distance zero .
as we will see , two graphons are equivalent if and only if the random families @xmath30 generated from these graphons have the same distribution ; see theorem [ prop11 ] below . to compare graphs and graphons ,
we embed a graph on @xmath16 vertices into the set of step functions over @xmath10 ^ 2 $ ] in the usual way by decomposing @xmath10 $ ] into adjacent intervals @xmath67 of lengths @xmath68 , and define a step function @xmath69 as the function which is equal to @xmath70 on @xmath71 if @xmath19 and @xmath20 are connected in @xmath72 , and equal to @xmath73 otherwise . extending @xmath69 to a function on @xmath52 by setting it to zero outside of @xmath10
^ 2 $ ] , we can then compare graphs to graphons on measure spaces of infinite mass , and in particular we get a notion of convergence in metric of a sequence of graphs @xmath74 to a graphon @xmath15 . in the classical theory of graph convergence ,
such a sequence will converge to the zero graphon whenever the sequence is sparse .
we resolve this difficulty by rescaling the input arguments of the step function @xmath69 so as to get a `` stretched graphon ''
@xmath75 satisfying @xmath76 .
equivalently , we may interpret @xmath77 as a graphon where the measure of the underlying measure space is rescaled .
see figure [ fig2 ] for an illustration , which also compares the rescaling in the current paper with the rescaling considered in @xcite .
we say that @xmath74 converges to a graphon @xmath15 ( with @xmath78 norm equal to 1 ) for the stretched cut metric iff @xmath79 .
graphons on @xmath1-finite measure spaces of infinite total measure may therefore be considered as limiting objects for sequences of sparse graphs , similarly as graphons on probability spaces are considered limits of dense graphs .
we prove that graphon processes converge to the generating graphon in the stretched cut metric ; see proposition [ prop4 ] .
we will also consider another family of random sparse graphs associated with a graphon @xmath15 over a @xmath1-finite measure space , and prove that these graphs are also converging for the stretched cut metric .
( -1.2500,-0.50000)(-0.75000,-1.3750 ) ; ( 0.50000,-0.87500)(-0.75000,-1.3750 ) ; ( 0,0)(0,0.93750 ) ; ( 0,0)(-2.6875,-1.0750 ) ; ( 0,0)(1.0750,-1.8813 ) ; ( 0,0.25000 ) node[right ] 1 ; ( 0.50000,-0.87500 ) node[right ] 1 ; ( -1.2500,-0.50000 ) node[above left ] 1 ; ( 0,0 ) circle [ radius=0.02 ] ; ( 0,0.25000 ) circle [ radius=0.02 ] ; ( 0,0.50000 ) circle [ radius=0.02 ] ; ( 0,0.75000 ) circle [ radius=0.02 ] ; ( -1.2500,-0.50000 ) circle [ radius=0.02 ] ; ( -2.5000,-1.0000 ) circle [ radius=0.02 ] ; ( 0.50000,-0.87500 ) circle [ radius=0.02 ] ; ( 1.0000,-1.7500 ) circle [ radius=0.02 ] ; ( -0.25000,0.15000)(-0.75000,-0.050000)(-0.65000,-0.22500)(-0.15000,-0.025000)cycle ; ( -0.25000,0.15000)(-0.75000,-0.050000)(-0.65000,-0.22500)(-0.15000,-0.025000)cycle ; ( -0.75000,-0.30000)(-0.75000,-0.050000)(-0.65000,-0.22500)(-0.65000,-0.47500)cycle ; ( -0.75000,-0.30000)(-0.75000,-0.050000)(-0.65000,-0.22500)(-0.65000,-0.47500)cycle ; ( -0.15000,-0.27500)(-0.15000,-0.025000)(-0.65000,-0.22500)(-0.65000,-0.47500)cycle ; ( -0.15000,-0.27500)(-0.15000,-0.025000)(-0.65000,-0.22500)(-0.65000,-0.47500)cycle ; ( 0.10000,0.075000)(-0.15000,-0.025000)(0.050000,-0.37500)(0.30000,-0.27500)cycle ; ( 0.10000,0.075000)(-0.15000,-0.025000)(0.050000,-0.37500)(0.30000,-0.27500)cycle ; ( -0.15000,-0.27500)(-0.15000,-0.025000)(0.050000,-0.37500)(0.050000,-0.62500)cycle ; ( -0.15000,-0.27500)(-0.15000,-0.025000)(0.050000,-0.37500)(0.050000,-0.62500)cycle ; ( 0.30000,-0.52500)(0.30000,-0.27500)(0.050000,-0.37500)(0.050000,-0.62500)cycle ; ( 0.30000,-0.52500)(0.30000,-0.27500)(0.050000,-0.37500)(0.050000,-0.62500)cycle ; ( -0.65000,-0.22500)(-1.1500,-0.42500)(-1.0500,-0.60000)(-0.55000,-0.40000)cycle ; ( -0.65000,-0.22500)(-1.1500,-0.42500)(-1.0500,-0.60000)(-0.55000,-0.40000)cycle ; ( -1.1500,-0.67500)(-1.1500,-0.42500)(-1.0500,-0.60000)(-1.0500,-0.85000)cycle ; ( -1.1500,-0.67500)(-1.1500,-0.42500)(-1.0500,-0.60000)(-1.0500,-0.85000)cycle ; ( -0.55000,-0.65000)(-0.55000,-0.40000)(-1.0500,-0.60000)(-1.0500,-0.85000)cycle ; ( -0.55000,-0.65000)(-0.55000,-0.40000)(-1.0500,-0.60000)(-1.0500,-0.85000)cycle ; ( 0.050000,-0.37500)(-0.20000,-0.47500)(0,-0.82500)(0.25000,-0.72500)cycle ; ( 0.050000,-0.37500)(-0.20000,-0.47500)(0,-0.82500)(0.25000,-0.72500)cycle ; ( -0.20000,-0.72500)(-0.20000,-0.47500)(0,-0.82500)(0,-1.0750)cycle ; ( -0.20000,-0.72500)(-0.20000,-0.47500)(0,-0.82500)(0,-1.0750)cycle ; ( 0.25000,-0.97500)(0.25000,-0.72500)(0,-0.82500)(0,-1.0750)cycle ; ( 0.25000,-0.97500)(0.25000,-0.72500)(0,-0.82500)(0,-1.0750)cycle ; ( -0.80625,1.4906 ) node canonical graphon ; ( 3.0125,-0.50000)(3.5125,-1.3750 ) ; ( 4.7625,-0.87500)(3.5125,-1.3750 ) ; ( 4.2625,0)(4.2625,0.93750 ) ; ( 4.2625,0)(1.5750,-1.0750 ) ; ( 4.2625,0)(5.3375,-1.8813 ) ; ( 4.7625,-0.87500 ) node[right ] 1 ; ( 3.0125,-0.50000 ) node[above left ] 1 ; ( 4.2625,0 ) circle [ radius=0.02 ] ; ( 4.2625,0.25000 ) circle [ radius=0.02 ] ; ( 4.2625,0.50000 ) circle [ radius=0.02 ] ; ( 4.2625,0.75000 ) circle [ radius=0.02 ] ; ( 3.0125,-0.50000 ) circle [ radius=0.02 ] ; ( 1.7625,-1.0000 ) circle [ radius=0.02 ] ; ( 4.7625,-0.87500 ) circle [ radius=0.02 ] ; ( 5.2625,-1.7500 ) circle [ radius=0.02 ] ; ( 4.0125,0.68125)(3.5125,0.48125)(3.6125,0.30625)(4.1125,0.50625)cycle ; ( 4.0125,0.68125)(3.5125,0.48125)(3.6125,0.30625)(4.1125,0.50625)cycle ; ( 3.5125,-0.30000)(3.5125,0.48125)(3.6125,0.30625)(3.6125,-0.47500)cycle ; ( 3.5125,-0.30000)(3.5125,0.48125)(3.6125,0.30625)(3.6125,-0.47500)cycle ; ( 4.1125,-0.27500)(4.1125,0.50625)(3.6125,0.30625)(3.6125,-0.47500)cycle ; ( 4.1125,-0.27500)(4.1125,0.50625)(3.6125,0.30625)(3.6125,-0.47500)cycle ; ( 4.3625,0.60625)(4.1125,0.50625)(4.3125,0.15625)(4.5625,0.25625)cycle ; ( 4.3625,0.60625)(4.1125,0.50625)(4.3125,0.15625)(4.5625,0.25625)cycle ; ( 4.1125,-0.27500)(4.1125,0.50625)(4.3125,0.15625)(4.3125,-0.62500)cycle ; ( 4.1125,-0.27500)(4.1125,0.50625)(4.3125,0.15625)(4.3125,-0.62500)cycle ; ( 4.5625,-0.52500)(4.5625,0.25625)(4.3125,0.15625)(4.3125,-0.62500)cycle ; ( 4.5625,-0.52500)(4.5625,0.25625)(4.3125,0.15625)(4.3125,-0.62500)cycle ; ( 3.6125,0.30625)(3.1125,0.10625)(3.2125,-0.068750)(3.7125,0.13125)cycle ; ( 3.6125,0.30625)(3.1125,0.10625)(3.2125,-0.068750)(3.7125,0.13125)cycle ; ( 3.1125,-0.67500)(3.1125,0.10625)(3.2125,-0.068750)(3.2125,-0.85000)cycle ; ( 3.1125,-0.67500)(3.1125,0.10625)(3.2125,-0.068750)(3.2125,-0.85000)cycle ; ( 3.7125,-0.65000)(3.7125,0.13125)(3.2125,-0.068750)(3.2125,-0.85000)cycle ; ( 3.7125,-0.65000)(3.7125,0.13125)(3.2125,-0.068750)(3.2125,-0.85000)cycle ; ( 4.3125,0.15625)(4.0625,0.056250)(4.2625,-0.29375)(4.5125,-0.19375)cycle ; ( 4.3125,0.15625)(4.0625,0.056250)(4.2625,-0.29375)(4.5125,-0.19375)cycle ; ( 4.0625,-0.72500)(4.0625,0.056250)(4.2625,-0.29375)(4.2625,-1.0750)cycle ; ( 4.0625,-0.72500)(4.0625,0.056250)(4.2625,-0.29375)(4.2625,-1.0750)cycle ; ( 4.5125,-0.97500)(4.5125,-0.19375)(4.2625,-0.29375)(4.2625,-1.0750)cycle ; ( 4.5125,-0.97500)(4.5125,-0.19375)(4.2625,-0.29375)(4.2625,-1.0750)cycle ; ( 3.4563,1.4906 ) node rescaled graphon ; ( 8.5250,0)(8.5250,0.93750 ) ; ( 8.5250,0)(5.8375,-1.0750 ) ; ( 8.5250,0)(9.6000,-1.8813 ) ; ( 8.5250,0.25000 ) node[right ] 1 ; ( 8.5250,0 ) circle [ radius=0.02 ] ; ( 8.5250,0.25000 ) circle [ radius=0.02 ] ; ( 8.5250,0.50000 ) circle [ radius=0.02 ] ; ( 8.5250,0.75000 ) circle [ radius=0.02 ] ; ( 7.2750,-0.50000 ) circle [ radius=0.02 ] ; ( 6.0250,-1.0000 ) circle [ radius=0.02 ] ; ( 9.0250,-0.87500 ) circle [ radius=0.02 ] ; ( 9.5250,-1.7500 ) circle [ radius=0.02 ] ; ( 8.0831,0.073223)(7.1992,-0.28033)(7.3760,-0.58969)(8.2598,-0.23614)cycle ; ( 8.0831,0.073223)(7.1992,-0.28033)(7.3760,-0.58969)(8.2598,-0.23614)cycle ; ( 7.1992,-0.53033)(7.1992,-0.28033)(7.3760,-0.58969)(7.3760,-0.83969)cycle ; ( 7.1992,-0.53033)(7.1992,-0.28033)(7.3760,-0.58969)(7.3760,-0.83969)cycle ; ( 8.2598,-0.48614)(8.2598,-0.23614)(7.3760,-0.58969)(7.3760,-0.83969)cycle ; ( 8.2598,-0.48614)(8.2598,-0.23614)(7.3760,-0.58969)(7.3760,-0.83969)cycle ; ( 8.7018,-0.059359)(8.2598,-0.23614)(8.6134,-0.85485)(9.0553,-0.67808)cycle ; ( 8.7018,-0.059359)(8.2598,-0.23614)(8.6134,-0.85485)(9.0553,-0.67808)cycle ; ( 8.2598,-0.48614)(8.2598,-0.23614)(8.6134,-0.85485)(8.6134,-1.1049)cycle ; ( 8.2598,-0.48614)(8.2598,-0.23614)(8.6134,-0.85485)(8.6134,-1.1049)cycle ; ( 9.0553,-0.92808)(9.0553,-0.67808)(8.6134,-0.85485)(8.6134,-1.1049)cycle ; ( 9.0553,-0.92808)(9.0553,-0.67808)(8.6134,-0.85485)(8.6134,-1.1049)cycle ; ( 7.3760,-0.58969)(6.4921,-0.94324)(6.6688,-1.2526)(7.5527,-0.89905)cycle ; ( 7.3760,-0.58969)(6.4921,-0.94324)(6.6688,-1.2526)(7.5527,-0.89905)cycle ; ( 6.4921,-1.1932)(6.4921,-0.94324)(6.6688,-1.2526)(6.6688,-1.5026)cycle ; ( 6.4921,-1.1932)(6.4921,-0.94324)(6.6688,-1.2526)(6.6688,-1.5026)cycle ; ( 7.5527,-1.1490)(7.5527,-0.89905)(6.6688,-1.2526)(6.6688,-1.5026)cycle ; ( 7.5527,-1.1490)(7.5527,-0.89905)(6.6688,-1.2526)(6.6688,-1.5026)cycle ; ( 8.6134,-0.85485)(8.1714,-1.0316)(8.5250,-1.6503)(8.9669,-1.4736)cycle ; ( 8.6134,-0.85485)(8.1714,-1.0316)(8.5250,-1.6503)(8.9669,-1.4736)cycle ; ( 8.1714,-1.2816)(8.1714,-1.0316)(8.5250,-1.6503)(8.5250,-1.9003)cycle ; ( 8.1714,-1.2816)(8.1714,-1.0316)(8.5250,-1.6503)(8.5250,-1.9003)cycle ; ( 8.9669,-1.7236)(8.9669,-1.4736)(8.5250,-1.6503)(8.5250,-1.9003)cycle ; ( 8.9669,-1.7236)(8.9669,-1.4736)(8.5250,-1.6503)(8.5250,-1.9003)cycle ; ( 7.7188,1.4906 ) node stretched graphon ; particular random graph models of special interest arise by considering certain classes of graphons @xmath15 . in @xcite
the authors consider graphons on the form @xmath80 ( with a slightly different definition on the diagonal , since they also allow for self - edges ) for certain decreasing functions @xmath81 . in this model
@xmath40 represents a sociability parameter of each vertex .
a multi - edge version of this model allows for an alternative sampling procedure to the one we present above ( * ? ? ?
* section 3 ) . in @xcite
a generalization of the model in @xcite to graphs with block structure is introduced . in this model
each node is associated to a type from a finite index set @xmath82:=\{1,\dots , k\}$ ] for some @xmath83 , in addition to its sociability parameter , such that the probability of two nodes connecting depends both on their type and their sociability .
more generally we can obtain sparse graphs with block structure by considering integrable functions @xmath84 $ ] for @xmath85 , and defining @xmath86\times { \mathbb{r}}_+$ ] and @xmath87 . as compared to the block model in @xcite , this allows for a more complex interaction within and between the blocks .
an alternative generalization of the stochastic block model to our setting is to consider infinitely many disjoint intervals @xmath88 for @xmath89 , and define @xmath90 for constants
@xmath91 $ ] . for the block model in @xcite and
our first generalization above ( with @xmath86\times { \mathbb{r}}_+$ ] ) , the degree distribution of the vertices within each block will typically be strongly inhomogeneous ; by contrast , in our second generalization above ( with infinitely many blocks ) , all vertices within the same block have the same connectivity probabilities , and hence the degree distribution will be more homogeneous . we can also model sparse graphs with mixed membership structure within our framework .
in this case we let @xmath92^{k}$ ] be the standard @xmath93-simplex , and define @xmath94 . for a vertex with feature
@xmath95 the first coordinate @xmath96 is a vector such that @xmath97 for @xmath98 $ ] describes the proportion of time the vertex is part of community @xmath98 $ ] , and the second coordinate @xmath40 describes the role of the vertex within the community ; for example , @xmath40 could be a sociability parameter .
for each @xmath99 $ ] let @xmath100 be a graphon describing the interactions between the communities @xmath101 and @xmath102 .
we define our mixed membership graphon @xmath12 by @xmath103 } { \widetilde}x^1_{k_1}{\widetilde}x^2_{k_2 } w_{k_1,k_2}(x^1,x^2).\ ] ] alternatively , we could define @xmath104 , which would provide a model where , for example , the sociability of a node varies depending on which community it is part of . in the classical setting of dense graphs ,
many papers only consider graphons defined on the unit square , instead of graphons on more general probability spaces .
this is justified by the fact that every graphon with a probability space as base space is equivalent to a graphon with base space @xmath10 $ ] .
the analogue in our setting would be graphons over @xmath0 equipped with the lebesgue measure .
as the examples in the preceding paragraphs illustrate , for certain random graph models it is more natural to consider another underlying measure space .
for example , each coordinate in some higher - dimensional space may correspond to a particular feature of the vertices , and changing the base space can disrupt certain properties of the graphon , such as smoothness conditions .
for this reason we consider graphons defined on general @xmath1-finite measure spaces in this paper .
however , we will prove that every graphon is equivalent to a graphon on @xmath0 equipped with the borel @xmath1-algebra and lebesgue measure , in the sense that their cut distance is zero ; see proposition [ prop7 ] . as stated before ,
our results then imply that they correspond to the same random graph model .
the set of @xmath10$]-valued graphons on probability spaces is compact for the cut metric .
for the possibly unbounded graphons studied in @xcite , which are real - valued and defined on probability spaces , compactness holds if we consider closed subsets of the space of graphons which are _ uniformly upper regular _
( see section [ sec : convergence - results ] for the definition ) . in our setting , where we look at graphons over spaces of possibly infinite measure , the analogous regularity condition is _ uniform regularity of tails _ if we restrict ourselves to , say , @xmath10$]-valued graphons .
in particular our results imply that a sequence of simple graphs with uniformly regular tails is subsequentially convergent , and conversely , that every convergent sequence of simple graphs has uniformly regular tails .
see theorem [ prop1 ] and the two corollaries following this theorem . in the setting of dense graphs , convergence for
the cut metric is equivalent to left convergence , meaning that subgraph densities converge .
this equivalence does not hold in our setting , or for the unbounded graphons in @xcite ; its failure is characteristic of sparse graphs , because deleting even a tiny fraction of the edges in a sparse graph can radically change the densities of larger subgraphs ( see the discussion in ( * ? ? ?
* section 2.9 ) ) .
however , randomly sampled graphs do satisfy a notion of left convergence ; see proposition [ prop3 ] .
as previously mentioned , the forthcoming paper @xcite will generalize and unify the theories and models presented in @xcite .
along with the introduction of a generalized model for sampling graphs and an alternative ( and weaker ) cut metric , we will prove a number of convergence properties of these graphs .
since the graphs in this paper are obtained as a special case of the graphs in @xcite , the mentioned convergence results also hold in our setting . in section [ sec2 ] we will state the main results of this paper , which will be proved in the subsequent sections . in section [ sec4 ]
we prove that the cut metric @xmath105 is well defined . in section [ sec : equiv ] we prove that any graphon is equivalent to a graphon with underlying measure space @xmath0 .
we also prove that under certain conditions on the underlying measure space we may define the cut metric @xmath105 in a number of equivalent ways . in section [ sec3 ]
we prove that certain random graph models derived from a graphon @xmath15 , including the graphon processes defined above , give graphs converging to @xmath15 for the cut metric .
we also prove that two graphons are equivalent ( i.e. , they have cut distance zero ) iff the corresponding graphon processes are equal in law . in section [ sec :
compact ] we prove that uniform regularity of tails is sufficient to guarantee subsequential metric convergence for a sequence of graphs ; conversely , we prove that every convergent sequence of graphs with non - negative edge weights has uniformly regular tails . in section [ sec8 ]
we prove some basic properties of sequences of graphs which are metric convergent , for example that metric convergence implies unbounded average degree if the number of edges diverge and the graph does not have too many isolated vertices , see proposition [ prop : unbdd_avg_deg ] below ; we also compare the notion of metric graph convergence in this paper to the one studied in @xcite . in section [ sec6 ]
we prove with reference to the kallenberg theorem for jointly exchangeable measures that graphon processes for integrable @xmath14 are uniquely characterized as exchangeable graph processes satisfying uniform tail regularity .
we also describe more general families of graphs that may be obtained from the kallenberg representation theorem if this regularity condition is not imposed . in section [ sec5 ]
we prove our results on left convergence of graphon processes .
after writing a first draft of this work , but a little over a month before completing the paper , we became aware of parallel , independent work by veitch and roy @xcite , who introduce a closely related model for exchangeable sparse graphs and interpret it with reference to the kallenberg theorem for exchangeable measures .
the random graph model studied in @xcite is ( up to minor differences ) the same as the graphon processes introduced in the current paper . aside from both introducing this model , the results of the two papers are essentially disjoint . while @xcite focuses on particular properties of the graphs in a graphon process ( in particular
, the expected number of edges and vertices , the degree distribution , and the existence of a giant component under certain assumptions on @xmath15 ) , our focus is graph convergence , the cut metric , and the question of when two different graphons lead to the same graphon process .
see also svante janson s subsequent paper @xcite expanding on the results of our paper , characterizing in particular when two graphons are equivalent , and proving additional compactness results for graphons over @xmath1-finite spaces .
we will work mainly with simple graphs , but we will allow the graphs to have weighted vertices and edges for some of our definitions and results .
we denote the vertex set of a graph @xmath72 by @xmath106 and the edge set of @xmath72 by @xmath107 . the sets @xmath106 and @xmath107 may be infinite , but we require them to be countable . if @xmath72 is weighted , with edge weights @xmath108 and vertex weights @xmath109 , we require the vertex weights to be non - negative , and we often ( but not always ) , require that @xmath110 ( note that @xmath111 is defined in such a way that for an unweighed graph , it is equal to @xmath112 , as opposed to the density , which is ill - defined if @xmath113 ) .
we define the edge density of a finite simple graph @xmath72 to be @xmath114 . letting @xmath115 denote
the positive integers , a sequence @xmath74 of simple , finite graphs will be called sparse if @xmath116 as @xmath117 , and dense if @xmath118 .
we start by recalling several notions from measure theory . for two measure spaces @xmath119 and @xmath120 , a measurable map @xmath121 is called _ measure - preserving _
iff for every @xmath122 we have @xmath123 .
two measure spaces @xmath2 and @xmath124 are called _ isomorphic _ iff there exists a bimeasurable , bijective , and measure - preserving map @xmath125 .
a borel measure space is defined as a measure space that is isomorphic to a borel subset of a complete separable metric space equipped with a borel measure . throughout most of this paper ,
we consider @xmath1-finite measure spaces , i.e. , spaces @xmath11 such that @xmath6 can be written as a finite or countably infinite union of sets @xmath126 with @xmath127 .
recall that a set @xmath128 is an atom if @xmath129 and if every measurable @xmath130 satisfies either @xmath131 or @xmath132 .
the measure space @xmath133 is atomless if it has no atoms .
every atomless @xmath1-finite borel space of infinite measure is isomorphic to @xmath134 , where @xmath135 is the borel @xmath1-algebra and @xmath136 is lebesgue measure ; for the convenience of the reader , we prove this as lemma [ prop9 ] below .
we also need the notion of a coupling , a concept well known for probability spaces : if @xmath137 is a measure space for @xmath138 and @xmath139 $ ] , we say that @xmath18 is a _ coupling _ of @xmath140 and @xmath141 if @xmath18 is a measure on @xmath142 with marginals @xmath140 and @xmath141 , i.e. , if @xmath143 for all @xmath144 and @xmath145 for all @xmath146 .
note that this definition of coupling is closely related to the definition of coupling of probability measures , which applies when @xmath147 . for probability spaces , it is easy to see that every pair of measures has a coupling ( for example , the product space of the two probability spaces )
. we prove the existence of a coupling for @xmath1-finite measure spaces in section [ sec4 ] , where this fact is stated as part of a more general lemma , lemma [ prop20 ] .
finally , we say that a measure space @xmath148 _ extends _ a measure space @xmath11 if @xmath149 , @xmath150 , and @xmath151 for all @xmath128 .
we say that @xmath133 is a _ restriction _ of @xmath152 .
we will work with the following definition of a graphon .
a _ graphon _ is a pair @xmath153 , where @xmath11 is a @xmath1-finite measure space satisfying @xmath154 and @xmath14 is a symmetric real - valued function @xmath155 that is measurable with respect to the product @xmath1-algebra @xmath156 and integrable with respect to @xmath157 .
we say that @xmath158 is a graphon over @xmath133 .
most literature on graphons define a graphon to be the function @xmath14 instead of the pair @xmath159 .
we have chosen the above definition since the underlying measure space will play an important role .
much literature on graphons requires @xmath14 to take values in @xmath10 $ ] , and some of our results will also be restricted to this case . the major difference between the above definition and the definition of a graphon in the existing literature ,
however , is that we allow the graphon to be defined on a measure space of possibly infinite measure , instead of a probability space.^2 $ ] . as a limit object for convergent graph sequences
it was introduced in @xcite , and graphons over general probability spaces were first studied in @xcite and @xcite . ]
one may relax the integrability condition for @xmath14 in the above definition such that the corresponding random graph model ( as defined in definition [ defn3 ] below ) still gives graphs with finitely many vertices and edges for each bounded time .
this more general definition is used in @xcite .
we work with the above definition since the majority of the analysis in this paper is related to convergence properties and graph limits , and our definition of the cut metric is most natural for integrable graphons .
an exception is the notion of subgraph density convergence in the corresponding random graph model , which we discuss in the more general setting of not necessarily integrable graphons ; see remark [ rem : w - notin - l1 ] below .
[ rmk : notl1 ] we will mainly study simple graphs in the current paper , in particular , graphs which do not have self - edges .
however , the theory can be generalized in a straightforward way to graphs with self - edges , in which case we would also impose an integrability condition for @xmath14 along its diagonal .
if @xmath160 , where @xmath6 is a borel subset of @xmath161 , @xmath135 is the borel @xmath1-algebra , and @xmath136 is lebesgue measure , we write @xmath162 to simplify notation . for example , we write @xmath163 instead of @xmath164 . for any measure space @xmath11 and integrable function @xmath165 , define the _ cut norm _ of @xmath14 over @xmath133 by @xmath166 if @xmath6 and/or @xmath18 is clear from the context we may write @xmath167 or @xmath168 to simplify notation .
let @xmath169 denote the @xmath78 norm .
for any graphon @xmath170 and any extension @xmath152 of @xmath133 we say that the graphon @xmath171 is _ the trivial extension of @xmath158 to @xmath152 _ if @xmath172 and @xmath173 .
for measure spaces @xmath119 and @xmath120 , a graphon @xmath170 , and a measurable map @xmath174 , we define the graphon @xmath175 by @xmath176 for @xmath177 .
we say that @xmath178 ( resp .
@xmath179 ) is a _
pullback _ of @xmath158 ( resp .
@xmath14 ) onto @xmath180 . for @xmath138 ,
let @xmath181 with @xmath182 be a graphon .
* if @xmath139 $ ] the _ cut metric _ @xmath105 and _ invariant @xmath183 metric _ @xmath184 are defined by @xmath185 where @xmath186 denotes projection for @xmath138 , and we take the infimum over all couplings @xmath18 of @xmath140 and @xmath141 . * if @xmath187 let @xmath188 be a @xmath1-finite measure space extending @xmath189 for @xmath138 such that @xmath190 $ ] .
let @xmath191 be the trivial extension of @xmath192 to @xmath193 , and define @xmath194 * we call two graphons @xmath59 and @xmath60 _ equivalent _ if @xmath195 . [ defn1 ] the following proposition will be proved in section [ sec4 ] . recall that a pseudometric on a set @xmath6 is a function from @xmath196 to @xmath0 which satisfies all the requirements of a metric , except that the distance between two different points might be zero
the metrics @xmath105 and @xmath184 given in definition [ defn1 ] are well defined ; in other words , under the assumptions of ( i ) there exists at least one coupling @xmath18 , and under the assumptions of ( ii ) the definitions of @xmath197 and @xmath198 do not depend on the choice of extensions @xmath199 .
furthermore , @xmath105 and @xmath184 are pseudometrics on the space of graphons .
[ prop8 ] while the two metrics @xmath105 and @xmath200 are not equivalent , a fact which is already well known from the theory of graph convergence for dense graphs , it turns out that the statement that two graphons have distance zero in the cut metric is equivalent to the same statement in the invariant @xmath78 metric .
this is the content of our next proposition , which will be proved in section [ sec : equiv ] .
( we will actually prove a generalization of this proposition involving an invariant version of the @xmath201 metric ; see proposition [ prop : dcut - dp ] . )
[ prop : dcut - d1 ] let @xmath59 and @xmath60 be graphons
. then @xmath195 if and only if @xmath202
. we will see in proposition [ prop10 ] that under additional assumptions on the underlying measure spaces @xmath57 and @xmath58 the cut metric can be defined equivalently in a number of other ways , giving , in particular , the equivalence of the definitions and in the case of two borel spaces of infinite mass .
note that the classical theory of graphons on probability spaces appears as a special case of the above definitions by taking @xmath133 to be a probability space .
our definition of the cut metric @xmath105 is equivalent to the standard definition for graphons on probability spaces ; see , for example , @xcite .
note that @xmath105 is not a true metric , only a pseudometric , but we call it a metric to be consistent with existing literature on graphons .
however , it is a metric on the set of equivalence classes as derived from the equivalence relation in definition [ defn1 ] ( iii ) .
we work with graphons defined on general @xmath1-finite measure spaces , rather than graphons on @xmath0 , since particular underlying spaces are more natural to consider for certain random graphs or networks .
however , the following proposition shows that every graphon is equivalent to a graphon over @xmath0 .
its proof follows a similar strategy as the proof of the analogous result for probability spaces in ( * ? ? ?
* theorem 3.2 ) and ( * ? ? ? * theorem 7.1 ) , and will be given in section [ sec : equiv ] .
for each graphon @xmath170 there exists a graphon @xmath203 such that @xmath204 .
[ prop7 ] to define graph convergence in the cut metric , one traditionally @xcite embeds the set of graphs into the set of graphons via the following map .
given any finite weighted graph @xmath72 we define the _ canonical graphon _
@xmath205)$ ] as follows .
let @xmath206 be an ordering of the vertices of @xmath72 .
for any @xmath207 let @xmath208 denote the weight of @xmath209 , for any @xmath210 let @xmath211 denote the weight of the edge @xmath212 , and for @xmath213 define @xmath214 . by rescaling the vertex weights if necessary we assume without loss of generality that @xmath215 .
if @xmath72 is simple all vertices have weight @xmath216 , and we define @xmath217 .
let @xmath67 be a partition of @xmath10 $ ] into adjacent intervals of lengths @xmath218 ( say the first one closed , and all others half open ) , and finally define @xmath69 by @xmath219 note that @xmath69 depends on the ordering of the vertices , but that different orderings give graphons with cut distance zero .
we define a sequence of weighted , finite graphs @xmath220 to be
_ _ sparse _ _ norm as in our definition , one may for example consider the fraction of edges with non - zero weight , either weighted by the vertex weights or not . in the current paper
we do not define what it means for a sequence of weighted graphs to be dense , since it is not immediate which definition is most natural , and since the focus of this paper is sparse graphs . ] if @xmath221 as @xmath222 .
note that this generalizes the definition we gave in the very beginning of section [ sec2 ] for simple graphs .
a sequence @xmath74 of graphs is then defined to be convergent in metric if @xmath223 is a cauchy sequence in the metric @xmath105 , and it is said to be convergent to a graphon @xmath15 if @xmath224 .
equivalently , one can define convergence of @xmath74 by identifying a weighted graph @xmath72 with the graphon @xmath225 , where @xmath226 consists of the vertex set @xmath106 equipped with the probability measure given by the weights @xmath227 ( or the uniform measure if @xmath72 has no vertex weights ) , and @xmath228 is the function that maps @xmath229 to @xmath108 . in the classical theory of graph convergence
a sequence of sparse graphs converges to the trivial graphon with @xmath230 .
this follows immediately from the fact that @xmath231 for sparse graphs . to address this problem ,
@xcite and @xcite considered the sequence of reweighted graphons @xmath232 , where @xmath233)$ ] with @xmath234 for any graph @xmath72 , and defined @xmath74 to be convergent iff @xmath232 is convergent .
the theory developed in the current paper considers a different rescaling , namely a rescaling of the _ arguments _ of the function @xmath69 , which , as explained after definition [ defn1a ] below , is equivalent to _ rescaling the measure _ of the underlying measurable space .
we define the _ stretched canonical graphon _
@xmath235 to be identical to @xmath236 except that we `` stretch '' the function @xmath69 to a function @xmath237 such that @xmath76 .
more precisely , @xmath238 , where @xmath239 note that in the case of a simple graph @xmath72 , each node in @xmath106 corresponds to an interval of length @xmath240 in the canonical graphon @xmath241 , while it corresponds to an interval of length @xmath242 in the stretched canonical graphon . it will sometimes be convenient to define stretched canonical graphons for graphs with infinitely many vertices ( but finitely .
] many edges ) .
our definition of @xmath69 makes no sense for simple graphs with infinitely many vertices , because they can not all be crammed into the unit interval
. instead , given a finite or countably infinite graph @xmath72 with vertex weights @xmath243 which do not necessarily sum to @xmath70 ( and may even sum to @xmath244 ) , we define a graphon @xmath245 by setting @xmath246 if @xmath247 , and @xmath248 if there exist no such pair @xmath249 , with @xmath250 being the interval @xmath251 where we assume the vertices of @xmath72 have been labelled @xmath252 , and @xmath253 for @xmath254 .
the stretched canonical graphon will then be defined as the graphon @xmath255 with @xmath256 a definition which can easily be seen to be equivalent to the previous one if @xmath72 is a finite graph .
alternatively , one can define a stretched graphon @xmath257 as a graphon over @xmath106 equipped with the measure @xmath258 , where @xmath259 for any @xmath260 . in the case
where @xmath261 , this graphon is obtained from the graphon representing @xmath72 by rescaling the probability measure @xmath262 to the measure @xmath258 , while the function @xmath263 with @xmath264 is left untouched .
note that any graphon with underlying measure space @xmath0 can be `` stretched '' in the same way as @xmath69 ; in other words , given any graphon @xmath163 we may define a graphon @xmath265 , where @xmath266 is defined to be the linear map such that @xmath267 , except when @xmath268 , in which case we define the stretched graphon to be @xmath73 .
but for graphons over general measure spaces , this rescaling is ill - defined .
instead , we consider a different , but related , notion of rescaling , by rescaling the measure of the underlying space , a notion which is the direct generalization of our definition of the stretched graphon @xmath257 .
* for two graphons @xmath181 with @xmath182 for @xmath138 , define the _ stretched cut metric _
@xmath269 by @xmath270 where @xmath271 with @xmath272 and @xmath273 .
( in the particular case where @xmath274 , we define @xmath275 . ) identifying @xmath72 with the graphon @xmath276 introduced above , this also defines the stretched distance between two graphs , or a graph and a graphon . * a sequence of graphs @xmath74 or graphons @xmath277
is called _ convergent in the stretched cut metric _ if they form a cauchy sequence for this metric ; they are called _ convergent to a graphon _ @xmath15 _ for the stretched cut metric _ if @xmath278 or @xmath279 , respectively .
[ defn1a ] note that for the case of graphons over @xmath0 , the above notion of convergence is equivalent to the one involving the stretched graphons @xmath280 of @xmath281 defined by @xmath282 to see this , just note that by the obvious coupling between @xmath136 and @xmath283 , where in this case @xmath283 is a constant multiple of lebesgue measure , we have @xmath284 , and hence @xmath285 . as a consequence , we have in particular that @xmath286 for any two graphs @xmath72 and @xmath287 .
note also that the stretched cut metric does not distinguish two graphs obtained from each other by deleting isolated vertices , in the sense that @xmath288 whenever @xmath72 is obtained from @xmath287 by removing a set of vertices that have degree @xmath73 in @xmath287 .
the following basic example illustrates the difference between the notions of convergence in the classical theory of graphons , the approach for sparse graphs taken in @xcite , and the approach of the current paper .
proposition [ prop13 ] below makes this comparison more general .
let @xmath289 .
for any @xmath290 let @xmath220 be an erds - rnyi graph on @xmath16 vertices with parameter @xmath291 ; i.e. , each two vertices of the graph are connected independently with probability @xmath291 .
let @xmath292 be a simple graph on @xmath16 vertices , such that @xmath293 vertices form a complete subgraph , and @xmath294 vertices are isolated .
both graph sequences are sparse , and hence their canonical graphons converge to the trivial graphon for which @xmath230 , i.e. , @xmath295 , where we let 0 denote the mentioned trivial graphon .
the sequence @xmath74 converges to @xmath296 ^ 2},[0,1])$ ] with the notion of convergence introduced in @xcite , but does not converge for @xmath269 .
the sequence @xmath297 converges to @xmath59 for the stretched cut metric , i.e. , @xmath298 , but it does not converge with the notion of convergence studied in @xcite .
[ ex1 ] the sequence @xmath297 defined above illustrates one of our motivations to introduce the stretched cut metric .
one might argue that this sequence of graphs should converge to the same limit as a sequence of complete graphs ; however , earlier theories for graph convergence are too sensitive to isolated vertices or vertices with very low degree to accommodate this .
the space of all @xmath10$]-valued graphons over @xmath10 $ ] is compact under the cut metric @xcite .
this implies that every sequence of simple graphs is subsequentially convergent to some graphon under @xmath105 , when we identify a graph @xmath72 with its canonical graphon @xmath241 .
our generalized definition of a graphon , along with the introduction of the stretched canonical graphon @xmath77 and the stretched cut metric @xmath269 , raises the question of whether a similar result holds in this setting .
we will see in theorem [ prop1 ] and corollary [ cor : subseq - conv ] below that the answer is yes , provided we restrict ourselves to uniformly bounded graphons and impose a suitable regularity condition ; see definition [ defn2 ] .
the sequence @xmath74 in example [ ex1 ] illustrates that we may not have subsequential convergence when this regularity condition is not satisfied .
let @xmath299 be a set of uniformly bounded graphons .
we say that @xmath299 has _ uniformly regular tails _ if for every @xmath300 we can find an @xmath301 such that for every @xmath302 with @xmath11 , there exists @xmath303 such that @xmath304 and @xmath305 .
a set @xmath306 of graphs has _ uniformly regular tails _ if @xmath307 for all @xmath308 and the corresponding set of stretched canonical graphons @xmath309 has uniformly regular tails .
[ defn2 ] in lemma [ prop26 ] we will prove that for a set of graphs with uniformly regular tails we may assume the sets @xmath310 in the above definition correspond to sets of vertices . note that if a collection @xmath299 of graphons has uniformly regular tails , then every collection of graphons which can be derived from @xmath299 by adding a finite number of the graphons to @xmath299 will still have uniformly regular tails .
in other words , if @xmath311 are such that the conditions of definition [ defn2 ] are satisfied for all but finitely many graphons in @xmath299 , then the collection @xmath299 has uniformly regular tails .
the following theorem shows that a necessary and sufficient condition for subsequential convergence is the existence of a subsequence with uniformly regular tails .
its proof will be given in section [ sec : compact ] .
[ prop1 ] every sequence @xmath277 of uniformly bounded graphons with uniformly regular tails converges subsequentially to some graphon @xmath15 for the cut metric @xmath105 .
moreover , if @xmath312 is non - negative then every @xmath105-cauchy sequence of uniformly bounded , non - negative graphons has uniformly regular tails .
two immediate corollaries of this proposition are the following results .
[ cor : complete ] the set of all @xmath10$]-valued graphons is complete for the cut metric @xmath105 , and hence also for @xmath269 .
[ cor : subseq - conv ] let @xmath74 be a sequence of finite graphs with non - negative , uniformly bounded edge weights such that @xmath313 for each @xmath290 .
then the following hold : 1 .
if @xmath74 has uniformly regular tails , then @xmath74 has a subsequence that converges to some graphon @xmath15 in the stretched cut metric .
2 . if @xmath74 is a @xmath314-cauchy sequence , then it has uniformly regular tails .
if @xmath74 is a @xmath314-cauchy sequence , then it converges to some graphon @xmath15 in the stretched cut metric .
the former of the above corollaries makes two assumptions : ( i ) the graphons are uniformly bounded , and ( ii ) the graphons are non - negative .
we remark that both of these conditions are necessary .
the set of all @xmath0-valued graphons is not complete for the cut metric @xmath105 ; see for example the proof ( * ? ? ?
* proposition 2.12(b ) ) for a counterexample .
the set of all @xmath315$]-valued graphons is also not complete , as the following example suggested to us by svante janson illustrates . for each @xmath290 let @xmath316 be a @xmath317-valued graphon supported in @xmath318 ^ 2 $ ] satisfying @xmath319 and @xmath320 , by defining @xmath321 to be an appropriately rescaled version of a graphon for a sufficiently large erds - rnyi random graph with edge density @xmath322 .
define @xmath323 by @xmath324 , and assume there is a graphon @xmath163 such that @xmath325 .
then we can find a sequence of measure - preserving transformations @xmath326 with @xmath327 , such that @xmath328 .
this implies that @xmath329 ^ 2}-v_k\|_\square=0 $ ] for each @xmath89 . since @xmath330 is a graphon associated with an erds - rnyi random graph it is a step graphon . for any intervals @xmath331 such that @xmath332 or @xmath333 we have @xmath334 ^ 2}-v_k){\mathbf{1}}_{i\times j}\|_\square=0 $ ] , so since @xmath14 takes values in @xmath315 $ ] we have @xmath335 . since @xmath336 ^ 2)}=1 $ ] this implies that @xmath337 ^ 2)}= 1 $ ] .
we have obtained a contradiction to the assumption that @xmath15 is a graphon , since for each @xmath290 we have @xmath338 ^ 2)}$ ] . for comparison ,
theorem 5.1 of @xcite shows that @xmath10$]-valued graphons on the probability space @xmath10 $ ] ( and hence any probability space ) form a compact metric space under @xmath105 .
compactness fails in our setting , because convergence requires uniformly regular tails , but completeness still holds .
our next result compares the theory of graph convergence in @xcite with the theory developed in this paper .
first we will define the _ rescaled cut metric _ @xmath339 .
a sequence of graphs is convergent in the sense defined in @xcite iff it converges for this metric . for two graphons @xmath340 and @xmath341 , where @xmath342 and @xmath343 are measure spaces of the same total measure , define @xmath344 , @xmath345 , and @xmath346 where we take the infimum over all measures @xmath18 on @xmath347 with marginals @xmath140 and @xmath141 , respectively .
for any graphs @xmath72 and @xmath287 we let @xmath348 and @xmath349 , respectively , denote the canonical graphons associated with @xmath72 and @xmath287 , and for any graphon @xmath158 we define @xmath350 for the notion of convergence studied in @xcite , _ uniform upper regularity _ plays a similar role to that of regularity of tails in the current paper .
more precisely , subsequential uniform upper regularity for a sequence of graphs or graphons defined over a probability space is equivalent to subsequential convergence to a graphon for the metric @xmath339 ( see ( * ? ? ?
* appendix c ) ) .
the primary conceptual difference is that the analogue of corollary [ cor : complete ] does not hold in the theory of @xcite .
we will now define what it means for a sequence of graphs or graphons to be uniformly upper regular .
a _ partition _ of a measurable space @xmath351 is a finite collection @xmath352 of disjoint elements of @xmath353 with union @xmath6 . for any graphon @xmath170 with @xmath119 and a partition @xmath352 of @xmath351 ,
define @xmath354 by averaging @xmath14 over the partitions .
more precisely , if @xmath355 for some @xmath356 , define @xmath357 , where @xmath358 a sequence @xmath359 of graphons @xmath360 over probability spaces @xmath361 is _ uniformly upper regular _ if there exists a function @xmath362 and a sequence @xmath363 of positive real numbers converging to zero , such that for every @xmath300 , @xmath290 , and partition @xmath352 of @xmath364 such that the @xmath365-measure of each part is at least @xmath366 , we have @xmath367 for any graph @xmath72 define the _ rescaled canonical graphon _ @xmath368)$ ] of @xmath72 to be equal to the canonical graphon @xmath348 of @xmath72 , except that we rescale the graphon such that @xmath369 .
more precisely , we define @xmath370)$ ] with @xmath371 .
we say that a sequence of graphs @xmath74 is uniformly upper regular if @xmath232 is uniformly upper regular , where we only consider partitions @xmath352 corresponding to partitions of @xmath372 , and we require every vertex of @xmath220 to have weight less than a fraction @xmath366 of the total weight of @xmath372 . the following proposition , which will be proved in section [ sec8 ] , illustrates the very different nature of the sparse graphs studied in @xcite and the graphs studied in this paper .
let @xmath74 be a sequence of simple graphs satisfying @xmath373 for each @xmath290 .
* if @xmath74 is sparse it can not both be uniformly upper regular and have uniformly regular tails ; hence it can not converge for both metrics @xmath269 and @xmath339 if it is sparse . *
assume @xmath74 is dense and has convergent edge density .
then @xmath74 is a cauchy sequence for @xmath269 iff it is a cauchy sequence for @xmath339 . if we do not assume convergence of the edge density , being a cauchy sequence for @xmath339 ( resp .
@xmath269 ) does not imply being a cauchy sequence for @xmath269 ( resp .
@xmath339 ) . [ prop13 ]
many natural properties of graphons are continuous under the cut metric , for example certain properties related to the degrees of the vertices . for graphons defined on probability spaces it was shown in ( * ? ? ?
* section 2.6 ) that the appropriately normalized degree distribution is continuous under the cut metric .
a similar result holds in our setting , but the normalization is slightly different : instead of the proportion of vertices whose degrees is at least @xmath136 times the average degree , we will consider a normalization in terms of the square root of the number of edges . given a graph @xmath72 and vertex @xmath374 , let @xmath375 denote the degree of @xmath376 , and given a graphon @xmath377 , define the analogous function @xmath378 by @xmath379 the following proposition is an immediate consequence of lemma [ lem : d_w ] . [
pro : deg - conv ] let @xmath380 be a sequence of graphons that converge to a graphon @xmath377 in the cut metric @xmath105 , and let @xmath381 be a point where the function @xmath382 is continuous . then @xmath383 .
in particular , @xmath384 whenever @xmath220 is a sequence of finite simple graphs that converge to a graphon @xmath385 in the stretched cut metric and @xmath386 is continuous at @xmath136 .
our final result in this section , which will be proved in section [ sec8 ] , is that graphs which converge for the stretched cut metric have unbounded average degree under certain assumptions , a result which also holds for graphs that converge under the rescaled cut metric ( * ? ? ?
* proposition c.15 ) .
let @xmath74 be a sequence of finite simple graphs such that the number of isolated vertices in @xmath220 is @xmath387 and such that @xmath388 . if there is a graphon @xmath15 such that @xmath389 , then @xmath74 has unbounded average degree .
[ prop : unbdd_avg_deg ] in this section we will present two random graph models associated with a given @xmath10$]-valued graphon @xmath170 with @xmath11 . we first define the graphon process already described in the introduction . sample a poisson random measure @xmath390 on @xmath391 with intensity given by @xmath392 ( see ( * ? ? ?
* chapter vi , theorem 2.15 ) ) , and identify @xmath390 with the collection of points @xmath393 at which @xmath390 has a point mass .
let @xmath49 be a graph with vertex set @xmath390 , such that for each pair of vertices @xmath394 and @xmath395 with @xmath396 , there is an edge between @xmath397 and @xmath398 with probability @xmath399 , independently for any two @xmath400 .
note that @xmath49 is a graph with countably infinitely many vertices , and that the set of edges is also countably infinite except if @xmath14 is equal to 0 almost everywhere .
for each @xmath38 let @xmath26 be the induced subgraph of @xmath49 consisting only of the vertices @xmath401 for which @xmath402 . finally define @xmath5 to be the induced subgraph of @xmath26 consisting only of the vertices having degree at least one . while @xmath26 is a graph on infinitely many vertices if @xmath37 , it has finitely many edges almost surely , and thus @xmath5 is a graph with finitely many vertices .
when @xmath403 the set of graphs @xmath404 considered above is identical in law to a sequence of @xmath158-random graphs as defined in @xcite for graphons over @xmath10 $ ] and in , for example , @xcite for graphons over general probability spaces . in the case
when @xmath37 it is primarily the graphs @xmath5 ( rather than @xmath26 ) which are of interest for applications , since the graphs @xmath26 have infinitely many ( isolated ) vertices .
but from a mathematical point of view , both turn out to be useful .
let @xmath170 be a @xmath10$]-valued graphon .
define @xmath405 ( resp .
@xmath406 ) to be a random family of graphs with the same law as the graphs @xmath25 ( resp .
@xmath30 ) defined above . * a random family of simple graphs is called a _
graphon process without isolated vertices _ generated by @xmath15 if it has the same law as @xmath407 , and it is called a _ graphon process with isolated vertices _ generated by @xmath15 if it has the same law as @xmath408 . * a random family @xmath409 of simple graphs
is called a _ graphon process _ if there exists a graphon @xmath15 such that after removal of all isolated vertices , @xmath410 has the same law as @xmath411 .
if @xmath412 is a graphon process , then we refer to @xmath5 as the _ graphon process at time @xmath50_. [ defn3 ] given a graphon @xmath413 one can define multiple other natural random graph models ; see below . however , the graph models of definition [ defn3 ] have one property which sets them apart from these models : exchangeability . to formulate this
, we first recall that a random measure @xmath414 in the first quadrant @xmath52 is jointly exchangeable iff for every @xmath415 , permutation @xmath1 of @xmath416 , and @xmath417 , @xmath418.\ ] ] here @xmath419 means equality in distribution , and , as usual , a random measure on @xmath52 is a measure drawn from some probability distribution over the set of all borel measures on @xmath52 , equipped with the minimal @xmath1-algebra for which the functions @xmath420 are measurable for all borel sets @xmath421 . consider a random family of graphs @xmath409 , viewed as a stochastic process taking values in the measurable space of simple graphs with finitely many edges equipped with the discrete @xmath1-algebra .
we will assume that @xmath410 is _ projective _ , meaning that if @xmath422 , then @xmath26 is an induced subgraph of @xmath423 .
note that it is not sufficient that @xmath26 is a subgraph of @xmath423 ; it must be an induced subgraph .
furthermore , we assume that a.s .
@xmath424 here @xmath425 means the limit as we approach @xmath50 from the right , and @xmath426 means the corresponding limit from the left . to discuss the meaning of the condition ,
define the birth time @xmath427 of a vertex @xmath428 as the infimum over all times @xmath50 such that @xmath429 .
the second condition then implies that all vertices are present at their birth time , while the third implies that at each time , at most one vertex is born .
labelling the vertices by their birth time in @xmath25 , we then define a random measure @xmath430 on @xmath52 by @xmath431 where each edge @xmath432 is counted twice so that the measure is symmetric .
it is not that hard to check that the measure @xmath414 is jointly exchangeable if @xmath410 is a graphon process with isolated vertices is more useful that the family @xmath433 : the latter only contains information about when a vertex first appeared in an edge in @xmath434 , and not information about when it was born . ]
generated from some graphon @xmath15 .
but it turns out that the converse is true as well , provided the sequence has uniformly regular tails .
the following theorem will be proved in section [ sec6 ] , and as in @xcite we will rely on the kallenberg theorem for jointly exchangeable measures ( * ? ? ?
* theorem 9.24 ) for this description .
veitch and roy @xcite have independently formulated and proved a similar theorem , except that their version does not include integrability of the graphon or uniform tail regularity of the sequence of random graphs . before stating our theorem , we note that given a locally finite symmetric measure @xmath414 that is a countable sum of off - diagonal , distinct atoms of weight one in the interior of @xmath52 , we can always find a projective family of simple graphs @xmath26 obeying the condition and the other assumptions we make above , and that up to vertices which stay isolated for all times , this family of graphs is uniquely determined by @xmath414 .
in other words , any projective family of countable simple graphs @xmath306 with finitely many edges at any given time can be transformed into one obeying the condition ( by letting the vertices appear in the graph @xmath5 exactly at the time they were born , and by merging vertices born at the same time ) , provided the measure @xmath435 has only point measures of weight one , and has no points on the diagonal and the coordinate axes .
let @xmath409 be a projective family of random simple graphs which satisfy , and define @xmath430 by .
then the following two conditions are equivalent : * the measure @xmath414 is a jointly exchangeable random measure and @xmath25 has uniformly regular tails .
* there is a @xmath0-valued random variable @xmath436 such that @xmath437 is a @xmath10$]-valued graphon almost surely , and such that conditioned on @xmath436 , @xmath25 ( modulo vertices that are isolated for all @xmath38 ) has the law of @xmath438 .
[ prop5 ] recall that we called two graphons equivalent if their distance in the cut metric @xmath105 is zero .
the following theorem shows that this notion of equivalence is the same as equivalence of the graphon process generated from two graphons , in the sense that the resulting random graphs have the same distribution .
we remark that in ( ii ) and ( iii ) the graphs @xmath439 and @xmath440 refer to graphs where the vertices have no other labels than possibly an index in @xmath416 ; in particular , we do not know the feature @xmath441 of each vertex , and for the graphs @xmath30 we do not know when all the vertices were born ( we only know when the various vertices were added to the graph ) .
also note that in ( ii ) we only identify the law of @xmath26 up to vertices which are isolated for all times ; it is clear that if we extend the underlying measure space @xmath133 and extend @xmath15 trivially to this measure space , the resulting graphon is equivalent to @xmath15 and the law of the graphs @xmath5 remains unchanged , while the law of @xmath26 might change due to additional isolated vertices .
the theorem will be proved in section [ sec3 ] .
for @xmath138 let @xmath181 be @xmath10$]-valued graphons , and let @xmath440 and @xmath442 be the graphon processes generated from @xmath192 with and without , respectively , isolated vertices .
then the following statements are equivalent : * @xmath443 ; * @xmath444 and @xmath445 are equal in law ( modulo vertices which are isolated for all times ) ; * @xmath446 and @xmath447 are equal in law .
[ prop11 ] as indicated before , in addition to the graphon processes defined above , there are several other natural random graph models generated from a graphon @xmath15 .
consider a sequence of probability measures @xmath448 on @xmath351 , and construct a sequence of random graphs @xmath220 as follows .
start with a single vertex @xmath449 with @xmath450 sampled from @xmath140 . in step @xmath16 , sample @xmath451 from @xmath365 , independently from all vertices and edges sampled so far , and for each @xmath452 , add an edge between @xmath453 and @xmath454 with probability @xmath455 , again independently for each @xmath19 ( and independently of all vertices and edges chosen before ) .
alternatively , sample an infinite sequence of independent features @xmath456 distributed according to @xmath457 , and let @xmath72 be the graph on infinitely many vertices with vertex set identified with @xmath458 , such that for any two @xmath459 there is an edge between @xmath460 and @xmath461 independently with probability @xmath462 . for
each @xmath290 let @xmath220 be the induced subgraph of @xmath72 consisting only of the vertices @xmath463 for which @xmath464 .
it was proven in @xcite that dense @xmath15-random graphs generated from graphons on probability spaces converge to @xmath15 .
the following theorem generalizes this to graphon processes , as well as for the alternative model defined in terms of a suitable sequence of measures @xmath365 . we will prove the theorem in section [ sec3 ] .
let @xmath170 with @xmath11 be a @xmath10$]-valued graphon . then the following hold : * almost surely @xmath465 and @xmath466 .
* let @xmath467 be the sequence of simple graphs generated from @xmath15 with arrival probabilities @xmath468 as described above , where we assume @xmath469 , @xmath470 and @xmath471 for all @xmath290 , and @xmath14 is not equal to 0 almost everywhere . then a.s.-@xmath472 if and only if @xmath473 .
[ prop4 ] left convergence is a notion of convergence where we consider subgraph counts of small test graphs .
existing literature defines left convergence both for dense graphs @xcite and for bounded degree graphs @xcite , with a different renormalization factor to adjust for the difference in edge density .
we will operate with a definition of subgraph density with an intermediary renormalization factor , to take into account that our graphon process satisfies @xmath474 .
for dense graphs our definition of left convergence coincides with the standard definition in the theory of dense graphs . for a simple graph @xmath475 and a simple graph @xmath72 define @xmath476 to be the number of adjacency preserving maps @xmath477 , i.e. , maps @xmath478 such that if @xmath479 , then @xmath480 , and define @xmath481 be the number of such maps which are injective .
define the _ rescaled homomorphism density _ @xmath482 and the _ rescaled injective homomorphism density _
@xmath483 of @xmath475 in @xmath72 by @xmath484 for any @xmath10$]-valued graphon @xmath170 we define the rescaled homomorphism density of @xmath475 in @xmath158 by @xmath485 note that in general , @xmath486 need not be finite .
take , for example , @xmath163 to be a graphon of the form @xmath487 where @xmath488 let @xmath489 .
then @xmath490 is in @xmath491 , but not in @xmath492 for any @xmath493 . thus if @xmath475 is a star with @xmath493 leaves , then @xmath494 .
proposition [ prop3](ii ) below , whose proof is based on lemma [ lem : h - bd ] , gives one criterion which guarantees that @xmath495 for all simple connected graphs @xmath475 .
a sequence @xmath74 is left convergent if its edge density is converging , and if for every simple connected graph @xmath475 with at least two vertices , the limit @xmath496 exists and is finite .
left convergence is defined similarly for a continuous - time family of graphs @xmath30 . for dense graphs left convergence
is equivalent to metric convergence @xcite .
this equivalence does not hold for our graphs , but convergence of subgraph densities ( possibly with an infinite limit ) does hold for graphon processes . the following proposition will be proved in section [ sec5 ] . *
if @xmath170 is a @xmath10$]-valued graphon and @xmath497 is a graphon process , then for every simple connected graph @xmath475 with at least two vertices , @xmath498\ ] ] almost surely . * in the setting of ( i ) , if @xmath499 is in @xmath201 for all @xmath500 , then @xmath501 for every simple connected graph @xmath475 with at least two vertices and @xmath502 almost surely , so in particular @xmath497 is left convergent . *
assume @xmath74 is a sequence of simple graphs with bounded degree such that @xmath503 and @xmath504 for all sufficiently large @xmath16 .
then @xmath74 is trivially left convergent , and @xmath505 for every connected @xmath475 for which @xmath506 . *
left convergence does not imply convergence for @xmath269 , and convergence for @xmath269 does not imply left convergence .
[ prop3 ] [ rem : w - notin - l1 ] while we stated the above proposition for graphons , i.e. , for the case when @xmath27 , the main input used in the proof , proposition [ prop2 ] below , does not require an integrable @xmath14 , but just the measurability of the function @xmath3 $ ] .
the main goal of this section is to prove proposition [ prop8 ] , which says that @xmath105 and @xmath200 are well defined and pseudometrics . in the course of our proof
, we will actually generalize this proposition , and show that it can be extended to the invariant @xmath201 metric @xmath507 , provided the two graphons are non - negative and in @xmath201 .
we start by defining the distance @xmath508 for two such graphons @xmath340 and @xmath509 over two spaces @xmath510 and @xmath511 of equal total measure , in which case we set @xmath512 where , as before , @xmath513 and @xmath514 are the projections from @xmath515 to @xmath516 and @xmath517 , respectively , and the infimum is over all couplings @xmath18 of @xmath140 and @xmath141 . if @xmath187 we define @xmath518 by trivially extending @xmath59 and @xmath60 to two graphons @xmath519 and @xmath520 , respectively , over measure spaces of equal measure , and defining @xmath521 , just as in definition [ defn1 ] ( ii ) .
[ prop : delta - p ] for @xmath138 , let @xmath181 be non - negative graphons over @xmath182 with @xmath522 for some @xmath523 . then @xmath508 is well defined .
in particular , @xmath508 does not depend on the choice of extensions @xmath524 and @xmath525 .
furthermore , @xmath507 is a pseudometric on the space of non - negative graphons in @xmath201
. we will prove proposition [ prop : delta - p ] at the same time as proposition [ prop8 ] .
we will also establish an estimate ( lemma [ prop16 ] ) saying that two graphons are close in the cut metric if we obtain one from the other by slightly modifying the measure of the underlying measure space .
finally we state and prove a lemma , lemma [ lem : d_w ] , that immediately implies proposition [ pro : deg - conv ] .
the following lemma will be used in the proof of propositions [ prop7 ] and [ prop10 ] .
the analogous result for probability spaces can be found in e.g. ( * ? ? ?
* theorem a.7 ) .
let @xmath11 be an atomless @xmath1-finite borel space .
then there exists a measure - preserving bijection of @xmath133 onto @xmath526 , where @xmath135 is the borel @xmath1-algebra and @xmath136 is lebesgue measure .
[ prop9 ] for @xmath403 , this follows from ( * ? ? ?
* theorem a.7 ) .
for @xmath37 we use that by the hypotheses of @xmath1-finiteness there exist disjoint sets @xmath527 for @xmath89 such that @xmath528 and @xmath529 for all @xmath89 . by ( * ? ? ?
* theorem a.7 ) , for every atomless borel probability space there is a measure - preserving bijection to @xmath10 $ ] equipped with the borel @xmath1-algebra and lebesgue measure ; hence , for each @xmath89 we can find measure - preserving bijections @xmath530\to [ 0,\mu(s_k))$ ] and @xmath531 $ ] .
it follows by considering the composed map @xmath532 that @xmath533 is isomorphic to @xmath534 .
the lemma follows by constructing a measure - preserving bijection from @xmath6 to @xmath0 where each set @xmath533 is mapped onto an half - open interval of length @xmath535 .
the first statement of proposition [ prop8 ] , i.e. , the existence of a coupling , follows directly from the following more general result . for @xmath536 let @xmath537 be a @xmath1-finite measure space such that @xmath538 $ ] .
let @xmath539 , and let @xmath540 be a measure on the product space @xmath541 , where @xmath542 is equipped with the induced @xmath1-algebra from @xmath533 .
assume the marginals @xmath543 and @xmath544 of @xmath540 are bounded above by @xmath545 and @xmath546 , respectively , and that either @xmath547 or @xmath548 for @xmath536 .
then there exists a coupling @xmath18 of @xmath57 and @xmath58 , such that @xmath549 .
[ prop20 ] first we consider the case when @xmath547 .
if @xmath550 we define @xmath18 to be proportional to the product measure of @xmath140 and @xmath141 . explicitly , for @xmath551 and @xmath552 , we set @xmath553 .
this clearly gives @xmath554 and @xmath555 , as required .
if @xmath556 , we consider partitions of @xmath516 and @xmath517 into disjoint sets of finite measure , with @xmath557 for @xmath536 .
let @xmath558 and @xmath559 be decompositions of @xmath560 into adjacent intervals of lengths @xmath561 and @xmath562 , respectively .
we then define a measure @xmath18 on @xmath563 by @xmath564 as a weighted sum of product measures , @xmath18 is a measure , and inserting @xmath565 or @xmath566 , one easily verifies that @xmath18 has marginals @xmath140 and @xmath141 .
this completes the proof of the lemma in the case that @xmath547 .
now we consider the general case .
decomposing @xmath567 and @xmath568 into disjoint sets of finite mass with respect to @xmath140 and @xmath141 , @xmath569 with @xmath570 , we define measures @xmath571 on @xmath533 for @xmath572 by @xmath573 \text { for all } b\in \cs_2,\qquad \\ { \widehat}\mu_1^{(2)}(a ) & = \frac 12 \mu_1(a\cap ( s_1\backslash d_1 ) ) + \sum_{i\geq 1}\bigl[\mu_1(a\cap d_i^1)-{\widetilde}\mu_1(a\cap d_i^1 ) \bigr]\text { for all } a\in \cs_1,\\ { \widehat}\mu_2^{(2)}(s_1)&=\frac 12\mu_2(b\cap(s_2\backslash d_2))\text { for all } b\in \cs_2 . \end{split}\ ] ] note that @xmath574 for @xmath575 by our assumption @xmath576 for @xmath536 . by the result for the case @xmath547 , for @xmath575
, we can find couplings @xmath577 of @xmath578 and @xmath579 on @xmath347 .
extending the measure @xmath540 to a measure on @xmath515 by assigning measure @xmath73 to all sets which have an empty intersection with @xmath541 , the measure @xmath580 has the appropriate marginals . to see that @xmath549 , we note that @xmath581 and @xmath582 , implying in particular that @xmath583 .
[ cor - to - prop20 ] for @xmath536 let @xmath537 be a @xmath1-finite measure space such that @xmath584 $ ] , and let @xmath18 be a coupling of @xmath140 and @xmath141 .
let @xmath539 be such that @xmath585 $ ] .
then there exists a coupling @xmath540 of @xmath140 and @xmath141 such that @xmath540 is supported on @xmath586 and @xmath587 on @xmath588 .
let @xmath589 be the restriction of @xmath18 to @xmath590 , let @xmath591 and @xmath592 be its marginals , and let @xmath593
. then @xmath594 and @xmath595 by the assumption of the corollary . in a similar way , @xmath596 .
with the help of the previous lemma , and considering the domains @xmath541 and @xmath597 separately , we then construct a coupling @xmath598 of @xmath200 and @xmath599 that has support on @xmath586 .
setting @xmath600 we obtain the statement of the corollary
. the following basic lemma will be used multiple times throughout this section . the analogous result for probability spaces
can be found in ( * ? ? ?
* lemma 6.4 ) .
let @xmath601 , let @xmath602 for @xmath138 be such that @xmath603 , and let @xmath340 , @xmath604 , and @xmath341 be graphons in @xmath201 . defining @xmath105 and @xmath507 as in definition [ defn1](i ) , we have @xmath605 and @xmath606 [ prop15 ] the second bound on @xmath197 is immediate , so the rest of the proof will consist of proving the first bound on @xmath197 as well as the bound on @xmath508 . let @xmath18 be a measure on @xmath607 with marginals @xmath140 and @xmath141 , respectively , and let @xmath608 denote projections for @xmath138 . since @xmath167 clearly satisfies the triangle inequality , @xmath609 the desired result follows by taking an infimum over all couplings .
the bound on @xmath508 follows in the same way from the triangle inequality for @xmath610 .
we only state the above lemma for the case when @xmath603 , since we have not yet proved that @xmath105 and @xmath507 are well defined otherwise . however , once we have proved this , it is a direct consequence of definition [ defn1](ii ) that the above lemma also holds when @xmath611 .
let @xmath351 be a measurable space , and consider a function @xmath612 .
then @xmath14 is a _ step function _ if there is some @xmath290 , disjoint sets @xmath613 satisfying @xmath127 for @xmath614 , and constants @xmath615 for @xmath616 such that @xmath617 [ defn4 ] note that in order for @xmath14 to be a step function it is not sufficient that it is simple , i.e. that it attains a finite number of values ; the sets on which the function is constant are required to be product sets .
the set of step functions is dense in @xmath78 ; hence lemma [ prop15 ] implies that every graphon can be approximated arbitrarily closely by a step function for the @xmath105 metric .
let @xmath601 , let @xmath618 and @xmath619 be graphons , and let @xmath620 and @xmath621 for finite index sets @xmath622 and @xmath623 such that @xmath624 for @xmath625 and @xmath626 for @xmath627 .
suppose @xmath628 and @xmath629 are step functions of the form @xmath630 where @xmath631 and @xmath632 are constants in @xmath161 .
let @xmath18 and @xmath589 be two coupling measures on @xmath347 , such that @xmath633 for all @xmath634
. then @xmath635 and @xmath636 .
[ prop23 ] for all @xmath637 , @xmath638 from the form of this expression and the definition of @xmath167 it is clear that we may assume there are sets @xmath639 such that for all @xmath640 , @xmath641 and @xmath642 and such that @xmath643 hence it follows from that @xmath635 if @xmath644 for all @xmath640 .
the proof for the @xmath201 metric follows from the fact that @xmath645 [ cor : cut - norm - dep - on - mu ] let @xmath601 and for @xmath536 , let @xmath646 with @xmath537 be graphons in @xmath201 . for @xmath536 , let @xmath647 and @xmath648 be extensions of @xmath649 with @xmath650 $ ] and @xmath651 $ ] , and let @xmath652 and @xmath653 be the trivial extensions of @xmath654 to @xmath655 and @xmath656
let @xmath540 and @xmath657 be couplings of @xmath543 and @xmath544 , and @xmath658 and @xmath659 , respectively . if @xmath540 and @xmath657 agree on @xmath347 , then @xmath660 and @xmath661 . by lemma [ prop15 ] and
the fact that step functions are dense in @xmath78 and in @xmath201 , it is sufficient to prove the corollary for step functions .
the corollary then follows from lemma [ prop23 ] by observing that for two sets @xmath662 and @xmath663 with finite measure @xmath664 and @xmath665 , the @xmath540 measure of sets of the form @xmath666 and @xmath667 can be expressed as @xmath668 and @xmath669 , respectively , implying that @xmath670 and @xmath671 depend only on the restriction of @xmath540 to @xmath347 , and similarly for @xmath672 and @xmath673 .
lemma [ prop23 ] is also used in the proof of the triangle inequality in the following lemma , which follows the same strategy as in the case of probability spaces ( * ? ? ?
* lemma 6.5 ) .
let @xmath601 .
for @xmath674 let @xmath675 with @xmath602 be a graphon in @xmath201 , such that @xmath676 $ ] . defining @xmath105 and @xmath507 as in definition [ defn1](i )
, we have @xmath677 [ prop24 ] by lemma [ prop15 ] and since step functions are dense in @xmath78 , we may assume that @xmath678 is a step function for @xmath674 .
let @xmath679 ( resp .
@xmath680 , @xmath681 ) be such that @xmath682 ( resp .
@xmath683 , @xmath684 ) is constant for all @xmath685 , and assume without loss of generality that @xmath686 for all @xmath687 . throughout the proof we abuse notation slightly and let @xmath688 denote projection onto @xmath689 from any space which is a product of @xmath689 and another space ; for example , @xmath513 denotes projection onto @xmath516 from @xmath690 , @xmath347 , and @xmath691 .
let @xmath300 , and let @xmath589 ( resp .
@xmath692 ) be a coupling measure on @xmath347 ( resp .
@xmath693 ) such that @xmath694 we define a measure @xmath18 on @xmath695 for any @xmath696 which is measurable for the product @xmath1-algebra by @xmath697 by a straightforward calculation ( see ( * ? ? ?
* lemma 6.5 ) ) the three mappings @xmath698 for @xmath699 are measure - preserving .
furthermore , if @xmath700 is the pushforward measure of @xmath18 for the projection @xmath701 , then @xmath702 by lemma [ prop23 ] and since @xmath703 is measure - preserving , @xmath704 hence , @xmath705 similarly , @xmath706 letting @xmath657 be the pushforward measure on @xmath691 of @xmath18 for the projection @xmath707 , we have @xmath708 since the cut norm @xmath167 clearly satisfies the triangle inequality , @xmath709 since @xmath710 was arbitrary this completes our proof for @xmath105 .
the proof for @xmath507 is identical .
let @xmath711 with @xmath602 be a graphon for @xmath138 , such that @xmath139 $ ] .
for @xmath138 let @xmath712 be an extension of @xmath189 , such that @xmath650 $ ] , and let @xmath713 be the trivial extension of @xmath281 to @xmath193
. then @xmath714 and @xmath715 , where @xmath105 and @xmath200 are as in definition [ defn1](i ) .
if @xmath716 and @xmath59 and @xmath60 are non - negative graphons in @xmath201 , then the result holds for @xmath507 as well .
[ prop21 ] we remark that the assumption of non - negativity is necessary for the lemma to hold when @xmath716 .
if for example @xmath717)$ ] and @xmath718)$ ] are graphons over @xmath10 $ ] , and if @xmath524 and @xmath525 are the trivial extensions to @xmath719 $ ] , then @xmath720 and @xmath721 .
we start with the proof for the cut metric . we will first prove the result for the case when @xmath722 and @xmath723 for @xmath138 and some @xmath89 , @xmath724 and @xmath725 are the associated discrete @xmath1-algebras , and @xmath726 for all @xmath727 and some @xmath728 .
first we will argue that @xmath729 by definition [ defn1](i ) it is sufficient to prove that for each coupling measure @xmath18 on @xmath196 we can define a coupling measure @xmath540 on @xmath730 such that @xmath731 .
but this is immediate , since we can define @xmath540 such that @xmath732 , and @xmath733 for @xmath734 .
next we will prove that @xmath735 again by definition [ defn1](i ) , it will be sufficient to prove that given any coupling measure @xmath540 on @xmath730 we can find a coupling measure @xmath18 on @xmath196 such that @xmath736 . by the following argument we may approximate @xmath670 arbitrarily well by replacing @xmath540 with a coupling measure which
is supported on @xmath737 , where @xmath738 for some sufficiently large @xmath83 .
indeed , by corollary [ cor - to - prop20 ] , given a coupling measure @xmath540 on @xmath730 and @xmath83 , we can define a measure @xmath657 supported on @xmath737 such that @xmath739 on @xmath737 .
it is easy to see from the construction of this measure in the proof of corollary [ cor - to - prop20 ] that when @xmath740 converges to infinity , the measure @xmath657 converges to @xmath540 when restricted to @xmath741 ( for example for the topology where we look at the maximum difference of the measure assigned to any set in @xmath741 ) .
therefore the corresponding cut norms also converge .
this shows that we may assume that @xmath540 is supported on @xmath737 for some @xmath83 when proving .
let @xmath700 be the restriction of @xmath540 to @xmath742 .
then @xmath743 where @xmath744 is the trivial extensions of @xmath745 to the measure space @xmath746 associated with @xmath747 .
we will prove that we may assume without loss of generality that @xmath700 corresponds to a permutation of @xmath748 . by choosing @xmath749 sufficiently large we can approximate @xmath750 arbitrarily well by replacing @xmath700 with a measure such that each element @xmath751 has a measure which is an integer multiple of @xmath752 ;
hence we may assume @xmath700 is on this form .
each such @xmath700 can be described in terms of a permutation @xmath753 of @xmath754 $ ] via @xmath755 .
let @xmath756)$ ] be the graphon such that each @xmath757 $ ] has measure @xmath752 , and such that @xmath758 for the measure - preserving map @xmath759\to[k]$ ] defined by @xmath760 . using proposition [ prop23 ] and the above observation on describing @xmath761 in terms of a permutation @xmath753 of @xmath754 $ ] we see that @xmath762 . upon replacing @xmath763 by @xmath764 throughout the proof
, we may assume that the measure @xmath700 is a permutation . to complete the proof it is therefore sufficient to consider some permutation @xmath765 of @xmath748 and prove that we can find a permutation @xmath1 of @xmath748 mapping @xmath6 to @xmath6 such that @xmath766 we modify the permutation @xmath765 step by step to obtain a permutation mapping @xmath6 to @xmath6 .
abusing notation slightly we let @xmath765 and @xmath1 denote the old and new , respectively , permutations in a single step . in each step choose @xmath767 and @xmath768 such that @xmath769 and @xmath770 ; if such @xmath771 do not exist we know that @xmath765 maps @xmath6 to @xmath6 . then define @xmath772 and @xmath773 , and for @xmath774 define @xmath775 .
we have @xmath776 by the following argument .
let @xmath777 be such that @xmath778 .
since @xmath779 ( implying that both @xmath780 and @xmath781 are trivial on @xmath782 and @xmath783 ) the following identity holds if we define @xmath784 or @xmath785 , and if we define @xmath786 or @xmath787 : @xmath788 in other words , @xmath789 is invariant under adding or removing @xmath790 from @xmath310 and/or @xmath791
. therefore we may assume without loss of generality that @xmath792 since if is not satisfied we may redefine @xmath310 and @xmath791 such that holds , and we still have @xmath793 .
the assumption implies that @xmath794 , which implies since we can obtain a permutation @xmath1 mapping @xmath6 to @xmath6 in finitely many steps as described above .
now we will prove the lemma for general graphons .
we will reduce the problem step by step to a problem with additional conditions on the measure spaces involved , until we have reduced the problem to the special case considered above .
first we show that we may assume @xmath189 and @xmath193 are non - atomic .
define @xmath795 $ ] and @xmath796 $ ] , let @xmath797 and @xmath798 be the corresponding atomless product measure spaces when @xmath10 $ ] is equipped with lebesgue measure , and let @xmath799 and @xmath800 be graphons such that @xmath801 and @xmath802 , where @xmath803 and @xmath804 are the projection maps on the first coordinates . by considering the natural coupling of @xmath805 and @xmath806 it is clear that @xmath807 .
it therefore follows from the triangle inequality that @xmath808 .
similarly , @xmath809 . in order to prove that @xmath810 it is therefore sufficient to prove that @xmath811 . since @xmath797 and @xmath798 are atomless and @xmath812 is a trivial extension of @xmath813 it is therefore sufficient to prove the lemma for atomless measure spaces .
next we will reduce the general case to the case when @xmath814 . if @xmath815 we extend @xmath193 to a space @xmath746 of infinite measure , and let @xmath816 be the trivial extension of @xmath817 to @xmath746 .
assuming we have proved the lemma for the case when the extended measure spaces have infinite measure , it follows that @xmath818 hence the lemma also holds for the case when @xmath815 .
next we prove that we may assume @xmath819 .
we proceed similarly as in the previous paragraph , and assume @xmath820 .
by lemma [ prop15 ] we may assume @xmath678 are supported on sets of finite measure , and we let @xmath821 be a restriction of @xmath189 such that @xmath822 and @xmath823 . since @xmath189 is non - atomic we may assume @xmath824 . define the graphon @xmath744 to be such that @xmath745 is the trivial extension of @xmath816 to @xmath189 .
assuming we have proved the lemma for the case when @xmath819 , it follows that @xmath818 hence the lemma also holds for the case when @xmath819 .
next we will prove that we may assume @xmath678 is a step function for @xmath138 , such that each step has the same measure @xmath825 .
step functions are dense in @xmath78 , and hence it is immediate from lemma [ prop15 ] that we may assume @xmath678 is a step function .
we may assume that the measure of each step is a rational multiple of @xmath826 ; if this is not the case we may adjust the steps slightly ( because @xmath189 is non - atomic , we can choose subsets of the steps of any desired measures , by exercise 2 from 41 in @xcite ) to obtain this .
assuming each step has a measure which is a rational multiple of @xmath826 we may subdivide each step such that each step obtains the same measure @xmath825 , again using the exercise from @xcite .
assume @xmath678 are step functions consisting of @xmath89 steps each having measure @xmath825 , and that @xmath819 and @xmath814 .
let @xmath827)$ ] ( resp .
@xmath828 ) be a graphon over @xmath829:=\{1,\dots , k\}$ ] ( resp .
@xmath416 ) equipped with the discrete @xmath1-algebra , such that each @xmath830 $ ] ( resp .
@xmath687 ) has measure @xmath831 , and such that @xmath832 ( resp .
@xmath833 ) for a measure - preserving map @xmath834 $ ] ( resp.@xmath835 ) . then @xmath836 and @xmath837 . by the special case we considered in the first paragraphs of the proof , @xmath838 . combining the above identities , our desired result @xmath839 follows . to prove the result for the metric @xmath507 , we follow the steps above .
the only place where the proof differs is in the proof of .
let @xmath1 , @xmath765 , and @xmath771 be as in the proof of .
we would like to show that @xmath840 writing both sides as a sum over @xmath841 we consider the following three cases separately : ( i ) @xmath842 , ( ii ) @xmath843 or @xmath844 , and ( iii ) @xmath845 . in case ( i ) the terms are identical on the left side and on the right side . for dealing with case ( ii )
it is sufficient to prove that for an arbitrary @xmath846 , @xmath847 this is equivalent to @xmath848 this inequality is obviously true if either @xmath849 or both @xmath780 and @xmath850 are non - negative . for case ( iii )
we need to show that @xmath851 writing out both sides we see that this is equivalent to @xmath852 which is again true if either @xmath849 or both @xmath780 and @xmath850 are non - negative .
the existence of a coupling follows from lemma [ prop20 ] with @xmath547 . to prove the next statement of the proposition , i.e. that the value of @xmath853 is independent of the extensions @xmath193 , we consider alternative extensions @xmath854 of @xmath189 for @xmath138 , and let @xmath855 denote the trivial extension of @xmath281 to @xmath746 . by lemma [ prop21 ]
it is sufficient to consider the case when @xmath856 , since if the extensions do not satisfy this property we can extend them to a space of infinite measure , and lemma [ prop21 ] shows that the cut norm is unchanged .
it is sufficient to prove that , given any coupling measure @xmath540 on @xmath857 , we can find a coupling measure @xmath657 on @xmath858 , such that @xmath859 by corollary [ cor : cut - norm - dep - on - mu ] , the left side of only depends on @xmath540 restricted to @xmath347 ; in a similar way , the right side only depends on @xmath657 restricted to @xmath347 .
we therefore can define an appropriate measure @xmath657 on @xmath860 by defining @xmath861 , and extending it to a coupling measure on @xmath858 by lemma [ prop20 ] ; this yields .
the function @xmath105 is clearly symmetric and non - negative . to prove that it is a pseudometric it is therefore sufficient to prove that it satisfies the triangle inequality .
this is immediate by lemma [ prop24 ] and the definition of @xmath105 as given in definition [ defn1](ii ) .
the proof for the metric @xmath200 follows exactly the same steps . using the statements of lemma [ prop21 ] , corollary [ cor : cut - norm - dep - on - mu ] , and lemma [ prop24 ] for @xmath716 , the above proof of proposition [ prop8 ] immediately generalizes to the invariant @xmath201 metric @xmath507 as long as the graphons in question are non - negative graphons in @xmath201 ( in addition to being in @xmath78 , as required by the definition of a graphon ) .
this proves proposition [ prop : delta - p ] . for two graphons @xmath12 and @xmath862 for which @xmath863 , it is immediate that @xmath864 .
the following lemma gives an analogous bound when @xmath14 and @xmath865 are defined on the same measurable space and @xmath866 , but the measures are not identical .
if @xmath18 and @xmath540 are two measures on the same measurable space @xmath867 and @xmath868 we write @xmath869 to mean that @xmath870 for every @xmath871 .
let @xmath12 with @xmath872 and @xmath873 with @xmath874 be graphons , and assume there is an @xmath875 such that @xmath876 .
then @xmath877 .
[ prop16 ] to distinguish between the graphons @xmath15 and @xmath878 we will write @xmath879 and @xmath148 , but recall throughout the proof that @xmath880 and @xmath881 . define @xmath882 , @xmath883 , and @xmath884 , and let @xmath885 be the disjoint union of @xmath133 and @xmath886 .
let @xmath887 be the trivial extension of @xmath15 to @xmath885 , and note that @xmath888 and @xmath878 are graphons over spaces of equal total measure . since @xmath889 it is sufficient to prove that @xmath890 .
let @xmath657 be the coupling measure on @xmath891 such that if @xmath892 , @xmath128 , and @xmath893 , then @xmath894 to complete the proof of the lemma it is sufficient to show that for all measurable sets @xmath895 , @xmath896 where @xmath897 ( resp .
@xmath898 ) is the projection onto the first ( resp .
second ) coordinate of @xmath891 .
let @xmath899 and @xmath900 be such that @xmath901 and @xmath902 . recall that since @xmath903
we may also view @xmath904 as sets in @xmath905 , and we denote these sets by @xmath906 , respectively . by
first using @xmath907 while @xmath657-almost surely @xmath908 on @xmath909 , then using @xmath910 , and then using @xmath880 and , we obtain the estimate : @xmath911 we close this section with a lemma that immediately implies proposition [ pro : deg - conv ] .
let @xmath912 and let @xmath377 and @xmath913 be two graphons with @xmath914 .
then @xmath915 for all @xmath916 .
[ lem : d_w ] since the trivial extension of a graphon @xmath15 only changes the measure of the set @xmath917 , we may assume without loss of generality that @xmath918 .
let @xmath513 and @xmath514 be the projections from @xmath919 onto the two coordinates , let @xmath920 , and let @xmath657 be a coupling of @xmath18 and @xmath589 such that @xmath921 by definition of the cut metric , this implies that @xmath922 for all @xmath923 . applying this bound for @xmath924 and @xmath925
, this implies that @xmath926 which in turn implies that @xmath927
as a consequence @xmath928 taking @xmath929 and using monotone convergence we obtain the first inequality in the statement of the lemma .
the second is proved in the same way .
let @xmath930 , and choose @xmath16 large enough so that @xmath931 .
by lemma [ lem : d_w ] , @xmath932 since @xmath933 is assumed to be continuous at @xmath136 , this gives @xmath934 the matching lower bound on the @xmath935 is proved in the same way .
in this section we will prove that every graphon is equivalent to a graphon over @xmath0 ( proposition [ prop7 ] ) , and prove that under certain assumptions on the underlying measure space of a graphon the cut metric can be defined in a number of equivalent ways ( proposition [ prop10 ] ) .
the first statement of the following lemma is a generalization of the analogous result for probability spaces ( * ? ? ?
* lemma 7.3 ) , and will be used to prove theorem [ prop1 ] and proposition [ prop7 ] .
we proceed similarly to the proof in @xcite , but in this case we also need to argue that underlying measure space of the constructed graphon is @xmath1-finite , and we include an additional result on atomless measure spaces . every graphon @xmath170 with @xmath119 is a pullback by a measure - preserving map of a graphon on some @xmath1-finite borel measure space . if @xmath133 is atomless , the @xmath1-finite borel space can be taken to be atomless as well .
[ prop6 ] let @xmath936 , and let @xmath937 be such that for each @xmath89 , we have @xmath527 , @xmath938 , @xmath529 , and @xmath939 . we claim that we can find a sequence of sets @xmath940 satisfying the following properties:(i ) if @xmath941 and @xmath942 , then @xmath14 is @xmath943-measurable , ( ii ) for all @xmath89 there exists @xmath944 such that @xmath945 , ( iii ) for each @xmath946 there exists a @xmath89 such that @xmath947 , and ( iv ) @xmath948 . a set @xmath949 satisfying ( i ) can be constructed by noting that each level set @xmath950 for @xmath951 is measurable with respect to @xmath952 for some countable set @xmath953 ( this follows , for example , by lemma 3.4 from @xcite ) . by adding the sets @xmath954 to @xmath949 we
obtain a collection of sets satisfying ( i ) and ( ii ) .
given a set @xmath955 satisfying ( i ) and ( ii ) , we can easily obtain an @xmath956 satisfying ( i)(iii ) by replacing each @xmath957 with the countable collection of sets @xmath958 .
finally , ( ii ) implies ( iv ) .
let @xmath959 be the cantor cube equipped with the product topology , and define @xmath960 by @xmath961 .
let @xmath962 be the pushforward measure of @xmath18 onto @xmath963 equipped with the borel @xmath1-algebra .
we claim that @xmath962 is a @xmath1-finite measure on @xmath963 . for each @xmath89 define @xmath964 and @xmath965 , where @xmath966 , and observe that all the subsets of @xmath963 just defined are measurable . the claim will follow if we can prove that ( a ) @xmath967 for each @xmath89 , and ( b ) @xmath968 .
property ( a ) is immediate since from the definition of @xmath962 , the fact @xmath969 , and the properties ( ii ) and ( iii ) of @xmath949 , which imply that @xmath970 . to prove ( b )
let @xmath971 .
we want to prove that @xmath972 for some @xmath89 . if @xmath973 we have @xmath974 , so @xmath972 for all @xmath89 . if @xmath975 for some @xmath441 , then @xmath976 for exactly one @xmath89 , so @xmath977 .
the argument in the following paragraph is similar to the proof of ( * ? ? ?
* lemma 7.3 ) , but we repeat it for completeness .
since the @xmath1-field on @xmath6 generated by @xmath478 equals @xmath978 , the @xmath1-field on @xmath196 generated by @xmath979 equals @xmath943 .
since @xmath14 is measurable with respect to @xmath943 we may use the doob - dynkin lemma ( see , for example , ( * ? ? ?
* lemma 1.13 ) ) to conclude that there exists a measurable function @xmath980 $ ] such that @xmath981 .
we may assume @xmath791 is symmetric upon replacing it by @xmath982 .
this completes the proof of the main assertion , since @xmath791 is a graphon on a @xmath1-finite borel measure space .
finally we will prove the last claim of the lemma , i.e. that if @xmath133 is atomless we may take @xmath962 to be atomless as well . to prove this claim it is sufficient to establish that the set @xmath956 in the above argument can be modified in such a way that @xmath983 for each @xmath984 .
given a collection of sets @xmath956 satisfying ( i)(iv ) above we define a new collection of sets @xmath955 as follows .
first define @xmath985 , and then define @xmath986 for @xmath987 inductively as follows .
for any @xmath987 and @xmath988 , let @xmath989 and @xmath990 be disjoint sets with union @xmath991 such that @xmath992 ; note that such sets @xmath993 and @xmath994 can be found since @xmath133 is atomless . then define @xmath995 , and finally define @xmath996 .
then @xmath955 is countable , satisfies ( i)(iv ) , and by defining the measure @xmath962 using @xmath955 instead of @xmath956 we have @xmath983 for every @xmath997 .
proceeding as above with @xmath955 instead of @xmath956 we get @xmath981 , where @xmath791 is a graphon over an atomless @xmath1-finite borel space .
proposition [ prop7 ] follows immediately from the following lemma , whose proof in turn follows a similar strategy as the proof of ( * ? ? ?
* theorem 7.1 ) .
[ lem : prop7 ] let @xmath170 be a graphon over an arbitrary @xmath1-finite space @xmath133 .
* there are two graphons @xmath203 and @xmath998 and measure - preserving maps @xmath999 and @xmath1000 such that @xmath1001 and @xmath1002 is the trivial extension of @xmath1003 from @xmath1004 to @xmath0 . *
if @xmath133 is a borel measure space , then we can find a measure - preserving map @xmath1005 such that @xmath1006 is a graphon over @xmath1004 equipped with the borel @xmath1-algebra and lebesgue measure . *
if @xmath133 is an atomless borel measure space , we may take @xmath1007 in ( ii ) to be an isomorphism between @xmath133 and @xmath1004 .
we start with the proof of ( ii ) and ( iii ) .
if @xmath133 is an atomless borel space the statement is immediate from lemma [ prop9 ] .
if @xmath133 has atoms , we define a graphon @xmath1008 where @xmath1009,\cs\times{\mathcal}b,\mu\times\lambda))$ ] is the product measure space and @xmath1010 , with @xmath1011\to s$ ] being the projection . since @xmath152 is an atomless borel space we may again use lemma [ prop9 ] , giving the existence of an isomorphism @xmath1012 such that @xmath1013 is a graphon over @xmath1014 equipped with lebesgue measure . observing that @xmath1015 , we obtain statement ( ii ) with @xmath1016 . to prove ( i ) we use that by lemma [ prop6 ] , @xmath15 can be expressed as @xmath1017 for a graphon @xmath888 on some borel space @xmath1018 and some measure - preserving map @xmath999 .
we then apply the just proven statement ( ii ) to the graphon @xmath888 , and define @xmath1019 to be the trivial extension of @xmath1020 from @xmath1004 to @xmath0 .
the statement of the proposition follows from lemma [ lem : prop7 ] by observing that @xmath1021 .
the following proposition provides equivalent definitions of the cut metric @xmath105 under certain assumption on the underlying measure spaces .
see ( * ? ? ?
* lemma 3.5 ) and ( * ? ? ?
* theorem 6.9 ) for analogous results for probability spaces . for @xmath63 let @xmath1022 with @xmath1023 be a graphon satisfying @xmath1024 .
then the following identities hold , and thus ( a)(e ) provide alternative definitions of @xmath105 under certain assumptions on the underlying measure spaces : * if @xmath1025 are borel spaces , then @xmath1026 , where we take the infimum over measure - preserving @xmath1027 for @xmath63 , where @xmath0 is equipped with the borel @xmath1-algebra and lebesgue measure . * if @xmath1025 are atomless borel spaces , then @xmath1028 , where we take the infimum over measure - preserving @xmath1029 . * if @xmath1025 are atomless borel spaces , then @xmath1028 , where we take the infimum over measure - preserving bijections @xmath1029 . * if @xmath1030 , then @xmath1031 , where we take the infimum over all interval permutations @xmath1032 ( i.e. , @xmath1032 maps @xmath250 to @xmath1033 for some permutation @xmath1 of the non - negative integers , and @xmath1034 $ ] for some @xmath415 ) . *
for @xmath63 let @xmath1035 be increasing sets satisfying @xmath1036 and @xmath1037 .
then @xmath1038 , where @xmath1039 and @xmath1040 is the restriction of @xmath1041 to @xmath1042 .
[ prop10 ] let @xmath1043 , @xmath1044 , @xmath1045 , and @xmath1046 denote the right sides of the equalities in ( a ) , ( b ) , ( c ) , and ( d ) , respectively . for @xmath63 fix some arbitrary sequence @xmath1035 satisfying @xmath1036 for all @xmath89 , @xmath1047 , and @xmath1037 . define @xmath1048 and @xmath1049 by @xmath1050 where @xmath1051 .
by lemma [ prop9 ] it is sufficient to consider the case @xmath1052 in ( b ) and ( c ) , since we can consider graphons @xmath1053 on @xmath0 , which satisfy @xmath1054 , by using measure - preserving transformations @xmath1055 . under this assumption
we have @xmath1056 , since we take the infimum over smaller and smaller sets of maps . by definition , @xmath1057 . to complete the proof of the proposition it is therefore sufficient to prove the following results : ( i ) @xmath1058 for general @xmath1-finite measure spaces @xmath1059 of infinite measure , ( ii ) @xmath1060 for @xmath1061 , and ( iii ) @xmath1062 .
we will start by proving ( i ) .
since @xmath1063 , lemma [ prop15 ] implies that it is sufficient to prove @xmath1058 for the case when @xmath1064 for some @xmath89 . under this assumption @xmath1065 for all @xmath1066 , and ( i ) follows .
now we will prove ( ii ) .
since @xmath1067 ^ 2}\|_1 = 0 $ ] by the dominated convergence theorem , as above we may assume by lemma [ prop15 ] that there is an @xmath301 such that @xmath14 is bounded and @xmath1068 ^ 2 $ ] for @xmath63 .
for @xmath63 define @xmath1069)$ ] , where @xmath1070 ^ 2}$ ] is a bounded graphon on @xmath1071 ^ 2 $ ] . by lemma [ prop21 ] and ( * ? ? ?
* lemma 3.5 ) ( or equivalently , ( * ? ? ?
* theorem 6.9 ) ) , @xmath1072 where @xmath1073 is an interval permutation of @xmath1071 $ ] and @xmath1074 is an interval permutation of @xmath0
. finally we will prove ( iii ) . by lemma [ lem : prop7 ]
there are measure - preserving maps @xmath1055 such that @xmath1075 .
the triangle inequality then implies that @xmath1076 .
since @xmath1077 and @xmath1078 are graphons over atomless borel spaces , it follows that @xmath1079 .
[ rem : alt - def - delta - p ] the proof of the above proposition clearly generalizes to the metric @xmath200 , the only additional ingredient being the analogue of ( * ? ? ?
* theorem 6.9 ) for the metric @xmath200 , ( * ? ? ? * remark 6.13 ) . using the results and proof techniques of appendix
a from @xcite instead of ( * ? ? ?
* theorem 6.9 ) , it can also be generalized to the metric @xmath507 for @xmath716 , again provided both graphons are non - negative and in @xmath201 .
we close this section by proving proposition [ prop : dcut - d1 ] .
in fact , we will prove a generalization of this proposition for the invariant @xmath201 metric @xmath507 .
the second statement of this proposition involves the distance @xmath518 of graphons @xmath1080 and @xmath1081 that are not necessarily non - negative , which means we do not have proposition [ prop : delta - p ] at our disposal to guarantee that @xmath507 is well defined .
we avoid this problem by defining @xmath507 as in proposition [ prop10 ] , i.e. , by setting @xmath1082 with the infimum going over measure - preserving bijections .
note that by remark [ rem : alt - def - delta - p ] , for non - negative graphons in @xmath201 , this definition is equivalent to the one given at the beginning of section [ sec4 ] .
[ prop : dcut - dp ] let @xmath601 , and let @xmath59 and @xmath60 be graphons in @xmath201 .
then * @xmath202 if and only if @xmath195 , and * if @xmath59 and @xmath60 are non - negative or graphons over @xmath0 , then @xmath1083 if and only if @xmath195 . for @xmath138 , let @xmath711 be a graphon over a borel space @xmath189 such that @xmath195 and @xmath603 .
then there exists a measure @xmath18 on @xmath347 such that * @xmath1084 , * the first ( resp .
second ) marginal of @xmath18 is dominated by @xmath140 ( resp .
@xmath141 ) , i.e. , for any @xmath1085 ( resp .
@xmath1086 ) we have @xmath1087 ( resp .
@xmath1088 ) , * if @xmath1085 is such that @xmath1089 then @xmath1090 , where @xmath1091 and the same property holds with the roles of @xmath59 and @xmath60 interchanged , and * in particular , if @xmath1092 with @xmath1093 as in ( iii ) , then @xmath18 is a coupling measure .
[ prop27 ] the analogous statement to proposition [ prop27 ] for graphons over probability spaces ( see , for example , ( * ? ? ? * theorem 6.16 ) ) states that when the underlying space is a borel probability space , the infimum in the definition of the cut distance using couplings is attained .
proposition [ prop27 ] says that the same result is true in our setting of @xmath1-finite measure spaces if we make two additional assumptions : ( a ) the cut distance between the graphons is zero , and ( b ) @xmath1094 for @xmath138 , where @xmath1093 is defined in part ( iii ) of the proposition .
we remark that both of these assumptions are necessary ; see the examples following this remark . in ( * ? ? ?
* theorem 5.3 ) , janson proves a related result , stating that if the cut distance of two graphons over @xmath1-finite borel spaces is zero , then there are trivial extensions of these graphons such that the extensions can be coupled so as to be equal almost everywhere .
proposition [ prop27 ] implies a similar result , namely that under this assumption , the restrictions of the two graphons to the sets @xmath1095 can be coupled so that they are equal a.e . to see this
, we note that by proposition [ prop : dcut - dp ] two graphons @xmath1096 with cut distance zero have distance zero in the metric @xmath200 , which in turn implies that @xmath1097 and @xmath1098 have distance zero in @xmath200 and hence in @xmath105 . by lemma [ lem : d_w ] , this in turn implies that @xmath1099 , which allows us to use proposition [ prop27 ] to deduce the claim .
[ rmk1 ] assumption ( a ) in remark [ rmk1 ] is necessary by the following counterexample , which illustrates that there are graphons @xmath1100 over @xmath0 such that @xmath1101 for all coupling measures @xmath18 .
let @xmath1080 be an arbitrary graphon such that @xmath628 is strictly positive everywhere , and let @xmath1081 be defined by @xmath1102 for @xmath1103 , @xmath1104 for @xmath1105 , and @xmath1106 otherwise .
first observe that @xmath1107 , since if @xmath1108 is defined by @xmath1109 for @xmath1110 $ ] , @xmath1111 for @xmath1112 $ ] , and @xmath1113 for @xmath1114 , then @xmath1115 then observe that @xmath1116 for all coupling measures @xmath18 , since if @xmath1117 and @xmath1118 $ ] then @xmath1119}w_2\,d\lambda\,d\lambda \right|>1 . \end{split}\ ] ] assumption ( b ) is necessary by the following counterexample , which illustrates that there are non - negative graphons @xmath1100 over @xmath0 such that @xmath195 and @xmath1120 for all coupling measures @xmath18 . letting @xmath59 and @xmath60 be defined as above , except that @xmath1106 for all @xmath1121 for which either @xmath1122 or @xmath1123 , we proceed as in case ( a ) to conclude that the graphons satisfy the desired property .
first we note that for @xmath147 , the proposition follows immediately from ( * ? ? ?
* theorem 6.16 ) , which in fact gives @xmath18 as a coupling of @xmath140 and @xmath141 .
the case @xmath1124 with @xmath1125 can be reduced to the case @xmath1126 by considering the graphons @xmath1127 where @xmath1128 is obtained from @xmath189 by multiplying the measures @xmath1129 by @xmath1130 , turning them into probability measures .
all that is left to consider is therefore the case @xmath556 .
next we argue that we may assume @xmath189 is atomless for @xmath138 . assuming the proposition has been proved for the case of atomless borel measure spaces , and given graphons @xmath281 over arbitrary borel measure spaces , we define graphons @xmath713 over the measure space @xmath193 defined as the product of @xmath189 and @xmath10 $ ] , such that @xmath1131 for the projection map @xmath1132\to s_{i}$ ] .
assume that @xmath540 is a measure on @xmath1133 such that the statements of the proposition hold for @xmath540 and @xmath1134 .
defining a measure @xmath18 for @xmath347 by letting @xmath18 be the pushforward of @xmath540 for the map @xmath1135 sending @xmath1136 , one easily checks that @xmath18 is a measure satisfying the conclusions of the proposition for @xmath59 and @xmath60 .
it follows that we may assume the spaces @xmath189 are atomless borel measure spaces , and by lemma [ prop9 ] we may assume that they are @xmath0 equipped with the lebesgue measure ; we will make these assumptions in the remainder of the proof .
in particular , we will no longer use the notation @xmath140 and @xmath141 from the proposition statement , since they are now both lebesgue measure @xmath136 ; it will be convenient to use the notation @xmath365 for other purposes .
consider a sequence of coupling measures @xmath448 such that @xmath1137 .
for any given @xmath301 , let @xmath1138 ^ 2}$ ] .
the measures @xmath1139 are not necessarily coupling measures , but their marginals are dominated by the lebesgue measure on @xmath1071 $ ] , and they satisfy @xmath1140 . furthermore , as a sequence of measures of uniformly bounded total mass over a compact metrizable space , they have a subsequence that converges in the weak topology ( * ? ? ?
* theorem 5.1 ) , i.e. , in the topology in which the integrals over all continuous functions on @xmath1071 ^ 2 $ ] converge .
let @xmath1141 be some subsequential limit , and note that as a limit of a sequence of measures having this property , the marginals of @xmath1141 are dominated by the lebesgue measure on @xmath1071 $ ] as well .
note also that by ( * ? ? ?
* theorem 2.8 ) , @xmath1142 converges weakly to @xmath1143 along any subsequence on which @xmath1139 converges weakly to @xmath1141 .
we will argue that @xmath1144 for all measurable subsets @xmath1145 ^ 2 $ ] .
indeed , given two such subsets and @xmath300 , let @xmath1146 be continuous functions over @xmath1071 ^ 2 $ ] such that @xmath1147 ^ 2}}\leq { \varepsilon}$ ] for @xmath138 , and let @xmath1148 ^ 2\to [ 0,1]$ ] be continuous functions such that @xmath1149 and @xmath1150 ( existence of appropriate functions @xmath1151 follows from , e.g. , ( * ? ? ?
* corollary 3.2.15 ) ) . using the fact that @xmath1152 and the fact that the marginals of @xmath1139 and @xmath1141 are dominated by the lebesgue measure on @xmath1071 $ ] , this allows us to conclude that for some sufficiently large @xmath16 chosen from the subsequence along which @xmath1139 converges
, @xmath1153 since @xmath710 was arbitrary , this proves . for each @xmath749
we let @xmath1141 be a measure as in the previous paragraph
. we may assume the subsequence along which @xmath1154 converges to @xmath1155 is a subsequence of the subsequence along which @xmath1139 converges to @xmath1141 .
this implies that @xmath1156 ^ 2}=\mu^m$ ] for all @xmath749 , so there is a measure @xmath18 on @xmath52 such that @xmath1157 ^ 2}=\mu^m$ ] .
furthermore , there is a subsequence of @xmath448 converging weakly to @xmath18 , such that for any @xmath749 the measures @xmath1158 ^ 2}$ ] converge weakly to @xmath1157 ^ 2}$ ] along this subsequence , and as a limit of measures with these properties , the measure @xmath18 satisfies ( ii ) , as well as @xmath1159 where , _ a priori _ , the supremum is over measurable , bounded subsets @xmath1160 .
but it is easy to see that if the supremum over these sets is zero , then the same holds for the supremum over all measurable subsets @xmath1161 ( use ( ii ) to conclude that the integrand is in @xmath78 , which means it can be approximated by functions over bounded subsets of @xmath1162 ) .
the property ( i ) of @xmath18 follows . it remains to prove that @xmath18 satisfies ( iii ) , since ( iv ) follows immediately from ( iii ) .
recall that the by definition of the measures @xmath365 , @xmath1163 where the supremum is over @xmath1164 and @xmath1165 .
fix any @xmath300 , and observe that for all @xmath1166 sufficiently large , @xmath1167 so @xmath1168 fix any @xmath1164 , and observe from the above that for @xmath740 sufficiently large @xmath1169 where @xmath1170 is the projection of @xmath1171}$ ] onto the first coordinate .
choose @xmath1172 such that @xmath1173 , where @xmath1174 is the space of continuous and compactly supported functions on @xmath52 ( the existence of such a function follows again from , e.g. , ( * ? ? ?
* corollary 3.2.15 ) ) .
for all @xmath1166 sufficiently large , @xmath1175 where @xmath1176 ( resp .
@xmath1177 ) is the projection of @xmath1178}$ ] ( resp .
@xmath18 ) onto the first coordinate .
next we claim that @xmath1179}$ ] converges weakly to @xmath1180}$ ] for any @xmath1181 . to see that , we need to show that for any continuous function @xmath1182\to{\mathbb{r}}_+$ ] the associated integral converges when @xmath222 . to this end , we approximate the function @xmath1183}$ ] ( with @xmath1184 for @xmath1185 ) by a function @xmath1186 , where @xmath1187
, @xmath1188 is a continuous function with support in @xmath1189 $ ] approximating @xmath1190 and satisfying @xmath1191 , and @xmath1192 $ ] is a continuous function with support in @xmath1193 $ ] approximating the indicator function of the set @xmath1193 $ ] . since
@xmath1194 and @xmath1195 can be chosen to be arbitrarily close approximations in the @xmath78 norm and the marginals of @xmath365 are given by lebesgue measure , this implies the claim .
therefore we can find @xmath1196 depending on @xmath740 , such that for all @xmath1197 @xmath1198 combining the above estimates and using the triangle inequality , we get that for sufficiently large @xmath740 and @xmath1197 , @xmath1199 since @xmath300 was arbitrary this implies that @xmath1200 since @xmath1201 is absolutely continuous with respect to @xmath136 , we know by the radon - nikodym theorem that there is a non - negative function @xmath1190 such that @xmath1202 .
the lebesgue differentiation theorem now says that @xmath1203 almost everywhere , which implies ( iii ) . since @xmath1204
, we only need to prove that @xmath195 implies @xmath202 in order to prove ( i ) .
assume first that the graphons are over @xmath0 , and let @xmath18 be as in proposition [ prop27 ] .
then @xmath1205 @xmath18-almost everywhere . for each @xmath290
let @xmath365 be some arbitrary coupling measure on @xmath347 such that @xmath1206 ^ 2}=\mu|_{[0,n]^2}$ ] .
then @xmath1207 , so @xmath202 . to obtain the result for graphons over general measure spaces we use lemma [ prop7 ] , the triangle inequality , and the fact that two graphons have distance zero for @xmath105 and @xmath200 if one is a pullback of the other . for ( ii ) with graphons over @xmath0 and @xmath105 defined in terms of measure - preserving transformations
, we will first prove that @xmath195 implies @xmath1083 .
this follows by the exact same argument as in the preceding paragraph , i.e. , by using proposition [ prop27 ] to construct a measure @xmath18 and coupling measures @xmath365 on @xmath347 .
now we will prove that @xmath1083 implies @xmath195 , still assuming the graphons are over @xmath0 and that @xmath507 is defined in terms of measure - preserving transformations . by part ( i )
it is sufficient to show that @xmath1083 implies @xmath202 .
fix @xmath300 and let @xmath1164 be such that @xmath1208 and @xmath1209 . by hlder s inequality , for any measure - preserving bijection @xmath1210 and @xmath1211 , @xmath1212 taking the infimum over @xmath478 we see that @xmath1213 . since @xmath710 was arbitrary , this shows that @xmath202 .
we get ( ii ) for non - negative graphons and @xmath507 defined in terms of couplings by using proposition [ prop10 ] and remark [ rem : alt - def - delta - p ] , and to move from graphons over @xmath0 to graphons over a general @xmath1-finite space we use lemma [ lem : prop7 ] .
the main goal of this section is to establish theorems [ prop11 ] and [ prop4 ] .
proposition [ prop2 ] will be used to prove left convergence of graphon processes in section [ sec7 ] .
it will also be applied in the proof of theorem [ prop4](i ) in this section , where we need to consider the normalized number of edges in a graphon process . a closely related result to the following proposition in the setting of graphons over probability spaces is ( * ? ? ?
* corollary 2.6 ) . however , the proofs are different , even if both rely on martingale techniques .
note that in the course of proving the below proposition we give an alternative proof of ( * ? ? ?
* theorem 5.3 ) for the special case of graphs with no self - edges .
recall from section [ sec7 ] that for a simple graph @xmath475 and a simple graph @xmath72 we let @xmath481 denote the number of injective adjacency preserving maps @xmath477
. let @xmath12 , where @xmath1214 $ ] is a symmetric , measurable ( but not necessarily integrable ) function , and @xmath11 is a @xmath1-finite measure space .
let @xmath475 be a simple graph with vertex set @xmath1215 for @xmath493 , such that @xmath475 has no isolated vertices .
where both sides should be read as elements of the extended non - negative reals @xmath1217 $ ] .
[ prop2 ] note that the proposition makes no integrability assumptions on @xmath14 .
all that is used is that @xmath14 is measurable . as a consequence
, the proposition can deal with situations where , say , the triangle density converges , even though @xmath1218 has infinitely many edges .
an extreme example of such a behavior can be obtained by taking @xmath163 to be the `` bipartite '' graphon defined by @xmath1219 , leading to a sequence of graphs @xmath1218 where every vertex has a.s . infinite degree , while all subgraph frequencies for graphs @xmath475 that are not bipartite converge to a finite value .
we first prove the proposition under the assumption that the right side of is finite . throughout the proof we let @xmath1220 be the graphon process generated by @xmath15 with isolated vertices .
note that the left side of is invariant under replacing @xmath1218 with @xmath1221 , because @xmath475 has no isolated vertices .
for each @xmath23 define @xmath1222 to be the left side of ( with @xmath1218 replaced by @xmath5 ) , i.e. , @xmath1223 as a first step , we will prove that for each @xmath23 , @xmath1224= \int_{s^k}\prod_{(i , j)\in e(f ) } w(x_i , x_j)\,dx_1\cdots\,dx_k .
\label{eq17}\ ] ] we may assume that @xmath403 , since we can write @xmath6 as a union of increasing sets @xmath1225 of finite measure for each @xmath356 , and by the monotone convergence theorem it is sufficient to establish with @xmath14 replaced by @xmath1226 , and with @xmath1222 defined in terms of graphs where we only consider vertices @xmath48 for which @xmath1227 . if @xmath1228 , then @xmath1229 . if @xmath1230 , then @xmath1231= \frac{1}{t^k } \frac{n!}{(n - k ) ! } \frac{1}{\mu(s)^k } \int_{s^k}\prod_{(i , j)\in e(f ) } w(x_i , x_j)\,dx_1\cdots\,dx_k,\ ] ] since we can form @xmath1232 ordered sets of size @xmath1233 from @xmath1234 , and the probability that a uniformly chosen injective map from @xmath1235 to @xmath1234 is a homomorphism , is given by @xmath1236 since @xmath1237 has the law of a poisson random variable with parameter @xmath1238 we can conclude that holds : @xmath1239 & = \sum_{n=0}^\infty { \mathbb{p}}(n = n ) { \mathbb{e}}[y_{-t}\,|\,n = n]\\ & = \sum_{n = k}^{\infty } \frac{(t\mu(s))^n}{n!}e^{-t\mu(s ) } \frac{1}{t^k } \frac{n!}{(n - k ) ! } \frac{1}{\mu(s)^k } \int_{s^k}\prod_{(i , j)\in e(f ) } w(x_i , x_j)\,dx_1\cdots\,dx_k\\ & = \int_{s^k}\prod_{(i , j)\in e(f ) } w(x_i , x_j)\,dx_1\cdots\,dx_k , \end{split}\ ] ] implying in particular that @xmath1240 is integrable for all @xmath1241 given that for now we assumed that the right side of is finite .
note that in particular this implies that @xmath1240 is a.s .
finite , even though @xmath14 may be such that the event that @xmath5 has infinitely many edges has non - zero probability .
let @xmath1242 be identical to @xmath5 , except that a vertex @xmath1243 is labelled only with @xmath40 .
in order words , conditioning on a realization of @xmath1242 is equivalent to conditioning on a realization of @xmath5 , except that the time the different vertices were born ( i.e. , the time they appeared in the graphon process @xmath25 ) is unknown .
note that since @xmath6 may have point masses multiple vertices of @xmath1242 may have the same label , but they are still considered to be different . for @xmath1244 , define @xmath1245 to be the @xmath1-algebra generated by @xmath1246 .
then @xmath1247 for @xmath1248 , so @xmath1249 is a filtration , and @xmath1250 is measurable with respect to @xmath1251 and hence @xmath1245 ; in other words , @xmath1252 is adapted to the filtration .
let @xmath1253 . given any @xmath1233 distinct vertices in @xmath1242 , the probability ( conditioned on @xmath1254 ) that all @xmath1233 vertices are also in @xmath1255
is given by @xmath1256 .
since @xmath1255 is an induced subgraph of @xmath1242 , it follows that @xmath1257&={\mathbb{e}}[y_{-s}\,|\ , ( { \widehat}g_{t'})_{t'\geq t}]\\ & = \frac{1}{s^k } \sum_{v_1,\dots , v_k\in v({\widehat}g_t ) } \prod_{(i , j)\in e(f)}{\mathbf{1}}_{(v_i , v_j)\in e(g_t ) } { \mathbb{p}}\big(v_1,\dots , v_k\in v({\widehat}g_s)\,|\,v_1,\dots , v_k\in v({\widehat}g_t)\big)\\ & = y_{-t } , \end{split}\ ] ] proving that @xmath1258 is a backwards martingale . by (
* theorem 7.18 ) the limit @xmath1259 exists almost surely .
since @xmath1260<\infty$ ] for all @xmath23 we know that a.s .
, @xmath1261 for all @xmath23 , which implies that a.s .
, @xmath1262 for all @xmath23
. therefore @xmath1263 has finitely many discontinuities in any bounded interval , and is left - continuous with limits from the right .
it follows that @xmath1264 ; i.e. , we do not need to take the limit along rationals . to complete the proof it is sufficient to prove that the limit @xmath1265 is equal to the right side of almost surely . to establish this
it is sufficient to prove that @xmath1265 is equal to a deterministic constant almost surely , since @xmath1263 is uniformly integrable ( * ? ? ?
* theorem 7.21 ) , which implies that @xmath1263 converges to @xmath1265 also in @xmath78 .
we will use the kolmogorov 0 - 1 law ( * ? ? ?
* theorem 1.1.2 ) to deduce this . for any
@xmath290 define @xmath1266 .
let @xmath1267 be the @xmath1-algebra generated by the set @xmath1268 and the edges between the vertex set @xmath1268 and the vertex set @xmath1269 .
since the randomness of the edges can be represented as an infinite sequence of independent uniform random variables , the @xmath1-algebras @xmath1267 can be considered independent even if the edges considered in @xmath1267 join vertices in @xmath1268 and @xmath1270 for @xmath1271 . in order to apply
the 0 - 1 law it is sufficient to prove that @xmath1265 is measurable with respect to the @xmath1-algebra generated by @xmath1272 for all @xmath1273 .
define @xmath1274 in the same way as @xmath1222 , except that instead of summing over vertices in @xmath1234 , we sum over vertices in @xmath1275 , where @xmath1276 .
since @xmath1274 is measurable with respect to the @xmath1-algebra generated by @xmath1272 , all we need to show is that for all @xmath1277 , a.s .
, @xmath1278 as @xmath1279 .
the difference between @xmath1222 and @xmath1274 can then be bounded by @xmath1280 } } } \prod_{(i , j)\in e(f)}{\mathbf{1}}_{(v_i , v_j)\in e(g_t ) } , \end{split}\ ] ] where @xmath1281 is the time label of @xmath209 .
conditioned on @xmath1282 , the probability that at least one of them has time label @xmath1283 is bounded by @xmath1284 .
continuing as in the proof of , we therefore obtain that @xmath1285\leq k\bigl(\frac { t_0}t\bigr)\int_{s^k}\prod_{(i , j)\in e(f ) } w(x_i , x_j)\,dx_1\cdots\,dx_k .
\label{eq17a}\ ] ] since the limit @xmath1286 exists , we can calculate @xmath1265 along any sequence , say the sequence @xmath1287 .
the bound combined with markov s inequality and the borel - cantelli lemma therefore implies that @xmath1288 , proving that @xmath1265 is measurable with respect to the @xmath1-algebra generated by @xmath1272 , as required for the application of the 0 - 1 law .
this completes the proof of the proposition under the assumption that the right side of is finite .
if the right side is infinite , we note that for any set @xmath991 of finite measure holds with @xmath1289 instead of @xmath14 on both the left side and the right side .
we can make the right side arbitrarily large by increasing @xmath991 .
the left side is monotone in @xmath991 , and therefore the limit inferior of the left side ( with @xmath14 , not @xmath1290 ) is larger than any fixed constant , and hence is equal @xmath244 .
since the result of the theorem is immediate for @xmath268 , we will assume throughout the proof that @xmath1291 .
since @xmath1292 by , it is enough to prove the statement for either @xmath25 or @xmath30 .
if @xmath6 has finite total mass , then , a.s . , @xmath1293 is finite for each fixed @xmath38 , and conditioned on the size of @xmath1293 , the graph @xmath26 is a @xmath15-random graph in the sense of the theory of dense graph convergence .
the results of @xcite imply that @xmath1294 and @xmath1295 where @xmath1296 is obtained from @xmath15 by normalizing the measure to a probability measure ( giving , in particular , @xmath1297 . ) combined with lemma [ prop16 ] , this implies that @xmath1298 when @xmath1299 .
if @xmath1300 , we use lemma [ lem : prop7 ] , and the observation that two graphons generate graphon processes with the same law if one graphon is a pullback of the other , to reduce the proof to the case @xmath1052 .
given @xmath1301 choose @xmath301 such that @xmath1302 ^ 2}\|_1<{\varepsilon}\|w\|_1 $ ] , and define @xmath1303 to be the graphon @xmath1304 ^ 2}$ ] over @xmath1071 $ ] , and @xmath1305 to be the induced subgraph of @xmath26 on the set of vertices @xmath1306 such that @xmath1307 .
define @xmath1308 with @xmath1309 . in the cut metric @xmath105
, the stretched graphon @xmath1310 then converges to @xmath1311 , again by the convergence of @xmath15-random graphs for @xmath15 defined on a probability space .
furthermore , by proposition [ prop2 ] applied to the graph @xmath475 consisting of a single edge , we have that a.s .
, the number of edges in @xmath1305 divided by @xmath1312 converges to @xmath1313 ^ 2}\|_1 $ ] , so in particular the time @xmath1314 where @xmath1305 has at least one edge is a.s.finite . for the rest of this proof
, we will always assume that @xmath1315 .
defining @xmath1316 to be the graph obtained from @xmath26 by removing all isolated vertices @xmath1306 from @xmath39 for which @xmath1317 , we note that by , it is sufficient to prove that @xmath1318 .
recall that each vertex @xmath1319 of @xmath1316 corresponds to an interval when we define the stretched canonical graphon @xmath1320 of @xmath1316 .
assume the intervals are ordered according the value of @xmath40 ; i.e. , if the vertices @xmath1319 and @xmath1321 satisfy @xmath1322 , then the interval corresponding to @xmath376 is to the left on the real line of the interval corresponding to @xmath1323 . noting that by our assumption @xmath1315 , there exists at last one vertex @xmath1319 in @xmath1316 such that @xmath1307 , we define the graphon @xmath1324 to be a `` stretched '' version of @xmath1325 such that the vertices @xmath1319 for which @xmath1326 $ ] correspond to the interval @xmath1327 $ ] .
in other words , @xmath1328 for some appropriately chosen constant @xmath1329 . to calculate @xmath1330 , we note that @xmath1331 with @xmath1332 and @xmath1333 with @xmath1334 , giving @xmath1335 . since @xmath1336 is an exponential random variable with expectation @xmath1337 , and @xmath1338 a.s . by proposition [ prop2 ] , we have that , a.s .
, @xmath1339 as @xmath1279 . by the triangle inequality , lemma [ prop15 ] , and the fact that @xmath1340 ^ 2}={\widetilde}{w}^{{{\widetilde}g}_t^m , s}$ ] , @xmath1341 ^ 2}-{\widetilde}{w}^{g_t',s}\|_1 + \delta_\square({\widetilde}{{{\mathcal}{w}}}^{g_t',s},{{\mathcal}{w}}^{g_t',s } ) .
\end{split } \label{eq24}\ ] ] the first term on the right side of is bounded by @xmath710 by assumption , and the second converges to zero as already discussed above .
the third term on the right side of is the product of @xmath1342 and the fraction of edges of @xmath5 for which at least one vertex @xmath48 satisfies @xmath1317 . by proposition [ prop2 ] ( applied with the random graphs @xmath1305 and @xmath26 and the same simple graph @xmath475 as above ) and @xmath1343 it follows that this term is less than @xmath1344 for all sufficiently large @xmath23 .
the fourth term on the right side of converges to zero by @xmath1343 and lemma [ prop16 ] .
since @xmath300 was arbitrary we can conclude that @xmath1345 .
first we will show that the condition @xmath1346 is necessary .
we will do the proof by contradiction , and assume @xmath1347 and a.s.-@xmath1348 .
we will obtain the contradiction by proving that with positive probability @xmath1349 for all @xmath290 ( which clearly contradicts a.s.-@xmath1350 ) . by rescaling the measure of @xmath133
we may assume without loss of generality that @xmath1351 .
furthermore , we assume that @xmath37 by extending @xmath133 and @xmath14 to a space of infinite measure .
note that the condition @xmath1352 will not hold after such an extension has been done , but we will not use this property in the proof . (
the property @xmath1352 is applied only in the second part of the proof , where we show that the condition @xmath473 is sufficient . )
first we will prove that there is a random @xmath1353 such that @xmath1354 ( up to interval permutations ) for all @xmath1355 .
since @xmath1356 is increasing , in order to do this it is sufficient to prove that @xmath1356 is bounded almost surely , and by monotone convergence , this in turn follows once we show that @xmath1357<\infty$ ] . letting @xmath1358 denote
the vertex added in step @xmath944 , and defining @xmath1359 , we obtain the desired result : @xmath1360 & = \sum_{1\leq i < j\leq n } { \mathbb{p}}[(v_i , v_j)\in e(g_n)]\\ & = \sum_{1\leq i < j\leq n } \frac{1}{\mu(s_i ) \mu(s_j ) } \| w{\mathbf{1}}_{s_i\times s_j}\|_1\\ & \leq \sum_{i',j'=1}^n \| w{\mathbf{1}}_{(s_{i'}\setminus s_{i'-1})\times ( s_{j'}\setminus s_{j'-1})}\|_1 \sum_{i\geq i',j\geq j ' } \frac{1}{\mu(s_i ) \mu(s_j)}\\ & \leq \|w\|_1 \left(\sum_{n=1}^\infty \mu(s_n)^{-1 } \right)^2\\ & < \infty . \end{split}\ ] ] since @xmath389 it follows that @xmath1361 .
we saw in the above paragraph that @xmath1361 a.s.for some random @xmath1353 .
therefore there is a deterministic step graphon @xmath1362 with values in @xmath1363 such that @xmath1364 .
since the set @xmath1365 has finite measure , by lemma [ lem : d_w ] , the set @xmath1366 has finite measure as well . after changing @xmath14 on a set of measure @xmath73
, we have @xmath1367 . note
also that by proposition [ prop : dcut - d1 ] we have @xmath1368 . for any @xmath290
the probability that a feature sampled from the measure @xmath365 is contained in @xmath991 is given by @xmath1369 .
hence the borel - cantelli lemma implies that finitely many vertices in @xmath1370 have a feature in @xmath991 .
therefore we can find a deterministic @xmath1371 such that @xmath1372 .
it follows that with uniformly positive probability conditioned on @xmath1373 , no edges are added to @xmath220 after time @xmath1374 . to conclude our proof ( i.e. , obtain a contradiction by proving that @xmath1349 with positive probability ) it is therefore sufficient to prove that @xmath1375 with positive probability .
we will do this by sampling a sequence of graphs @xmath1376 from @xmath1377 which is close in law to @xmath74 , and use that @xmath1378 with positive probability since @xmath1379 is zero on a certain subdomain since the graphs we consider have no loops .
( note that our approach would not have worked if we allowed for loops ; if , for example , @xmath1380 ^ 2}\equiv 1 $ ] and @xmath1381 $ ] we would have had @xmath1382 . )
let @xmath300 , and recalling that @xmath1383 , choose a coupling measure @xmath540 on @xmath1384 such that @xmath1385 . by using @xmath540 we can sample two coupled sequences of graphs @xmath1386 and @xmath1387 , such that the two sequences have a law which is close in total variation distance
, @xmath1386 has the law of the graphs in the statement of the theorem , and @xmath1388 is sampled similarly as @xmath74 but with @xmath1377 instead of @xmath15 . more precisely , for each @xmath1389 we sample @xmath1390 from the probability measure @xmath1391 , we let @xmath40 ( resp .
@xmath1392 ) be the feature of the @xmath16th vertex of @xmath220 ( resp .
@xmath1393 ) , and by using that @xmath1385 we can couple @xmath1394 and @xmath1388 such that for each @xmath1395 for which @xmath1396 we have @xmath1397 hence the total variation distance between the laws of @xmath1398 and @xmath1388 is bounded by @xmath1399 . since we can make this distance arbitrarily small by decreasing @xmath710 , in order to complete our proof it is sufficient to prove that @xmath1400 with a uniformly positive probability for all coupling measures @xmath540 .
write @xmath1401 , such that @xmath1402 correspond to the steps of the step function @xmath1379 , with , say , @xmath1403 corresponding to the set of all @xmath40 such that @xmath1404 . for any choice of @xmath540 we can find a @xmath1405 such that @xmath1406 .
therefore there is a uniformly positive probability that all the vertices of @xmath1407 have a feature in @xmath1408 . on this event
we have @xmath1400 , since @xmath1409 as the graphs we consider are simple ( i.e. , they do not have loops ) .
this completes our proof that the condition @xmath473 is necessary .
now we will prove that the condition @xmath473 is sufficient to guarantee that a.s.-@xmath1350 .
we will couple @xmath74 to a graphon process @xmath25 with isolated vertices .
fix @xmath300 , and choose @xmath1353 sufficiently large such that @xmath1410 .
sample @xmath25 , and independently from @xmath25 , sample @xmath1411 as described in the statement of the theorem .
define @xmath1412 inductively as follows @xmath1413 note that @xmath1414 a.s . as @xmath1415 , because the increments @xmath1416 are independent and exponentially distributed with mean @xmath1417 , and @xmath1418 by assumption .
for @xmath1419 let @xmath1393 be the induced subgraph of @xmath1420 whose vertex set is @xmath1421 for each @xmath1422 let @xmath220 be the union of @xmath1423 and @xmath1393 , such that the edge set of @xmath220 is given by @xmath1424 in addition to independently sampled edges between the vertices of @xmath1423 and the vertices of @xmath1425 , such that the probability of connecting vertices with features @xmath40 and @xmath41 is @xmath42 , and such that @xmath1426 is an induced subgraph of @xmath220 .
it is immediate that @xmath74 has the same law as the sequence of graphs @xmath74 described in the statement of the theorem .
we will prove that @xmath1427 and @xmath1428 for all large @xmath290 .
this is sufficient to complete the proof of the theorem , since @xmath300 was arbitrary , since @xmath1429 by part ( i ) of the theorem , and since @xmath1430 as @xmath222 and @xmath1431 by the following argument .
define @xmath1432 and @xmath1433 for @xmath1434 ; i.e. , @xmath1435 is a stretched version of @xmath1436 defined such that each vertex of @xmath220 corresponds to an interval of length @xmath1437
. then each vertex corresponds to an interval of length @xmath1437 both for @xmath1435 and for @xmath1438 , so by ordering the vertices appropriately when defining the graphons we have @xmath1439 . for sufficiently small @xmath300 and large @xmath290 we have @xmath1440 , and
hence lemma [ prop16 ] implies that @xmath1441 for all sufficiently small @xmath300 and sufficiently large @xmath290 . combining the above estimates we get that for all sufficiently small @xmath300 and sufficiently large @xmath290
, @xmath1442 first we prove that conditioned on almost any realization of @xmath1423 , @xmath1427 as @xmath1443 .
note that @xmath1444 consists of the edges in @xmath1445 , plus independently sampled edges between @xmath1446 and @xmath1447 . since @xmath1448
, we overcount the latter if we independently sample one edge for each @xmath1449 and @xmath1450 , with the probability of an edge between @xmath376 and @xmath1323 given by @xmath14 evaluated at the features of @xmath376 and @xmath1323 . defining @xmath1451 to be the number of edges between @xmath1449 and @xmath1452 obtained in this way , we thus have @xmath1453 by proposition [ prop2 ] applied with @xmath475 being the simple connected graph with two vertices , @xmath1454 . in order to prove that @xmath1427 it is therefore sufficient to prove that , conditioned on almost any realization of @xmath1423 , each vertex @xmath1449 satisfies @xmath1455 for all sufficiently large @xmath16 and some @xmath1456 depending on the feature of the vertex .
condition on a realization of @xmath1423 such that @xmath1457 for all @xmath441 such that @xmath40 is the feature of some vertex in @xmath1423 .
we will prove that if @xmath441 is the feature of @xmath1449 then a.s.@xmath1458 which is sufficient to imply the existence of an appropriate constant @xmath1459
. the convergence result follows by noting that @xmath1460 is a backwards martingale with expectation @xmath1461 , which is left - continuous with right limits at each @xmath1241 ; see the proof of proposition [ prop2 ] for a very similar argument .
hence the kolmogorov 0 - 1 law implies . we can conclude that @xmath1427
now we prove @xmath1428 .
let @xmath1462 be the induced subgraph of @xmath1420 corresponding to vertices with feature in @xmath1463
. then @xmath1464 by applying proposition [ prop2 ] to each of the graphs @xmath1462 and @xmath1465 , and with @xmath475 being the simple connected graph on two vertices , it follows that @xmath1466 assume that ( i ) holds , i.e. , @xmath195 .
we will prove that ( ii ) and ( iii ) also hold .
it is sufficient to prove that ( ii ) holds , since ( ii ) implies ( iii ) .
we first consider the case when @xmath1467 for @xmath138 , where @xmath1468 . recall that by proposition [ pro : deg - conv ] we have @xmath1469 , so by restricting the graphon @xmath281 to the space @xmath250 for @xmath138 we obtain two graphons with cut distance zero over spaces of finite and equal measure . by definition of @xmath250 ,
almost surely no vertices of @xmath25 will be isolated for all times , and it follows by ( * ? ? ?
* theorem 8.10 ) ( which refers to @xcite for original proofs ) that ( ii ) holds .
next we consider the case where @xmath1470 .
we may assume @xmath1471 , since replacing the graphon @xmath281 by its restriction to @xmath1472 amounts to removing vertices which are isolated for all times .
part ( i ) of proposition [ prop27 ] now implies that we can find a measure @xmath18 such that @xmath1473 @xmath18-almost everywhere . by the assumption @xmath1471 , part ( iii ) of the proposition
implies that @xmath18 is a coupling measure . sampling a graphon process from @xmath281 may be done by associating the vertex set with a poisson point process on @xmath1474 with intensity @xmath1475 , such that each @xmath1476 is associated with a vertex with feature @xmath1477 appearing at time @xmath50 .
now we will prove that ( ii ) or ( iii ) imply ( i ) .
we will only show that ( ii ) implies ( i ) , since we can prove that ( iii ) implies ( i ) by the exact same argument .
we assume ( ii ) holds , and couple @xmath1478 and @xmath1479 such that @xmath1480 for all @xmath1481 . by theorem [ prop4](i ) we know that @xmath1482 . since @xmath1483 for all @xmath38 it follows by the triangle inequality that @xmath195 , so ( i ) holds .
in this section we will establish theorem [ prop1 ] .
let @xmath277 and @xmath1484 be two sequences of graphons , with @xmath1485 , @xmath1486 , @xmath1487 , and @xmath1488 , such that there are measure - preserving transformations @xmath1489 for which @xmath1490 .
furthermore , assume that either ( i ) @xmath1491 is a bimeasurable bijection , or ( ii ) @xmath1492 $ ] , where @xmath10 $ ] is equipped with lebesgue measure , and @xmath1489 is the projection map
. then @xmath277 has uniformly regular tails iff @xmath1484 has uniformly regular tails .
[ prop25 ] let @xmath1493 be a sequence of positive real numbers converging to zero , such that @xmath1494 for all @xmath290 .
first assume @xmath1495 has uniformly regular tails .
given any @xmath300 let @xmath301 be such that for all @xmath290 we can find @xmath1496 satisfying @xmath1497 and @xmath1498 .
define @xmath1499 . since @xmath1491 is measure - preserving , @xmath1500 . by
first using @xmath1494 ( which implies that @xmath1501 ) and then using that @xmath1491 is measure - preserving we get @xmath1502 the right side is less than @xmath710 for all sufficiently large @xmath290
. therefore @xmath277 has uniformly regular tails .
next assume @xmath277 has uniformly regular tails .
we consider the two cases ( i ) and ( ii ) separately . in case
( i ) it is immediate from the above result that @xmath1495 has uniformly regular tails , since @xmath1503
. now consider case ( ii ) .
given any @xmath300 let @xmath301 be such that for all @xmath290 we can find @xmath1504 satisfying @xmath1505 and @xmath1506 .
define @xmath1507 by @xmath1508 and define @xmath1509 .
note that @xmath1507 is a measurable set since @xmath1510 is measurable .
then @xmath1497 , since @xmath1511 next we will argue that @xmath1512 if @xmath1513 it holds by the definition of @xmath1507 that @xmath1514 which implies by @xmath1515 using that @xmath1491 is measure - preserving , the triangle inequality , that @xmath1516 and the estimate we get @xmath1517 the right side is less than @xmath710 for all sufficiently large @xmath290 , and thus @xmath1495 has uniformly regular tails
. first we will prove that every @xmath105-cauchy sequence has uniformly regular tails .
let @xmath359 with @xmath360 be a @xmath105-cauchy sequence of graphons , i.e. , @xmath1518 . by lemma [ prop25 ]
we may assume without loss of generality that @xmath1519 is atomless for all @xmath290 . by lemmas [ prop6 ] and [ prop9 ] we can find graphons @xmath1520 and measure - preserving maps @xmath1521 such that @xmath1522 . since @xmath1523
, @xmath1495 is a cauchy sequence . given any
@xmath300 choose @xmath1353 such that @xmath1524 for all @xmath1355 .
for each @xmath1525 let @xmath1526 be such that @xmath1527 ^ 2}\|_1<{\varepsilon}/3 $ ] , and define @xmath1528 .
to prove that @xmath1495 has uniformly regular tails it is sufficient to prove that for each @xmath1355 we can find a borel - measurable set @xmath1529 such that @xmath1530 we can clearly find an appropriate set @xmath1531 for @xmath1532 ; indeed , we can find a set @xmath1533 such that the second bound holds with @xmath1534 instead of @xmath710 . by proposition [ prop10](c )
we can find measure - preserving bijections @xmath1535 such that @xmath1536 for all @xmath1355 .
define @xmath1537 , and note that @xmath1538 observing that for non - negative graphons the cut norm is equal to the @xmath78 norm , this gives that is satisfied and @xmath1495 has uniformly regular tails .
defining @xmath1539 , we have @xmath1540 and @xmath1541 .
hence @xmath277 has uniformly regular tails .
now we will prove that uniform regularity of tails implies subsequential convergence for @xmath105 .
we consider some sequence of graphons @xmath277 with uniformly regular tails , and will prove that the sequence is subsequentially convergent for @xmath105 towards some graphon @xmath15 . by lemma [ prop25 ]
we may assume without loss of generality that @xmath1542 is atomless for all @xmath290 , and by trivially extending @xmath312 to a graphon over a space of infinite total mass if needed , we may assume that @xmath1543 .
recall the definition of a partition of a measurable space , which was given as part of the discussion before the statement of proposition [ prop13 ] .
we will prove that we can find increasing sequences @xmath1544 and @xmath1545 with values in @xmath416 , such that for each @xmath1546 there is a partition @xmath1547 of @xmath364 and a graphon @xmath1548 such that the following hold : * @xmath1549 , where @xmath1550 and @xmath1551 for @xmath1552 .
* @xmath1553 for all @xmath290 . * for each @xmath1554 the value of @xmath1555 on @xmath1556 is constant and equal to the value of @xmath1557 on @xmath1558 . on the complement of @xmath1559
^ 2 $ ] , we have @xmath1560 . *
the partition @xmath1561 refines the partition @xmath1547 .
we number the elements of the partition @xmath1561 to be consistent with the refinement .
more precisely , defining @xmath1562 to be the ratio of the partition sizes in the two partitions , we have @xmath1563 for every @xmath19 with @xmath1564 . partitions @xmath1547 and graphons @xmath1565 satisfying ( i)(iv ) exist by the following argument . by the assumption of uniformly regular tails , for each @xmath89 we can find an @xmath1566 such that for appropriate sets @xmath1567 satisfying @xmath1550 we have @xmath1568 for all @xmath290 . by lemmas [ prop6 ] and [ prop9 ] , for each @xmath1569
the graphon @xmath1570 is a pullback of a graphon @xmath1571)$ ] by a measure - preserving transformation @xmath1572 . by applying szemerdi regularity for equitable partitions to @xmath1573 ( see , for example , lemma 3.3 in @xcite )
we can find appropriate @xmath1574 and partitions @xmath1575 of @xmath1559 $ ] such that @xmath1576 .
then the pullback of @xmath1577 along @xmath1572 equals @xmath1578 for an appropriate partition of @xmath364 satisfying ( i ) , and @xmath1579 define @xmath1580 as described in ( iii ) , and note that all requirements ( i)(iv ) are satisfied since @xmath1581 . by compactness , for each @xmath89 there exists a step function @xmath1582 $ ] ( with support in @xmath1559 ^ 2 $ ] ) such that @xmath1583 converges pointwise and in @xmath78 along a subsequence towards @xmath1584 .
we may assume the subsequence along which @xmath1585 converges is contained in the subsequence along which @xmath1583 converges .
note that for each @xmath1554 the function @xmath1584 is constant on @xmath1586\times[i_2 - 1,i_2])m_k / m_k$ ] .
furthermore , observe that if @xmath1587 and @xmath1066 , the value of @xmath1584 at @xmath1586\times[i_2 - 1,i_2])m_k / m_k$ ] is equal to the average of @xmath1588 over this set .
define the graphon @xmath1589 by @xmath1590 .
choose @xmath1591 , and then choose @xmath1592 such that @xmath1593 for all @xmath1594 .
let @xmath1595 be a uniformly random point in @xmath1596 ^ 2 $ ] . by the observations in the preceding paragraph
@xmath1597 is a martingale .
hence the martingale convergence theorem implies that the limit @xmath1598 exists a.s .
since @xmath1599 was arbitrary it follows that there is a set @xmath1600 of measure zero outside of which @xmath1601 converges pointwise .
define the graphon @xmath1602 as follows .
for any @xmath1603 define @xmath1604 , and for any @xmath1605 define @xmath1606 . since the functions @xmath1584 are uniformly bounded
, martingale convergence also implies that @xmath1607 ^ 2}$ ] converges to @xmath1608 ^ 2}$ ] in @xmath78 for each @xmath1609 .
next we will show that @xmath1610 .
since @xmath1611 ^ 2}\|_1=0 $ ] for each @xmath1612 it is sufficient to prove that @xmath1613 ^ 2}\|_1<1/(3\ell)$ ] for all @xmath1614 for which @xmath1615 .
this follows by fatou s lemma and the inequality @xmath1616 ^ 2 } |w_{n , k}| & = \int_{[0,m_k]^2\setminus[0,m_\ell]^2 } |w_{n , k}| = \int_{(s_n\setminus i_{n , k}^0)^2 \setminus ( s_n\setminus i_{n,\ell}^0)^2 } |w_{n}|_{{\mathcal}p_{n , k}}\\ & = \int_{(s_n\setminus i_{n , k}^0)^2 \setminus ( s_n\setminus i_{n,\ell}^0)^2 } |w_{n}| <
1/(3\ell ) .
\end{split}\ ] ] by the result of the preceding paragraph @xmath1617 , and we conclude the proof by applying the triangle inequality @xmath1618
in this section we will establish propositions [ prop13 ] and [ prop : unbdd_avg_deg ] .
first we prove a lemma saying that for a set of graphs with uniformly regular tails we may assume the sets @xmath310 in definition [ defn2 ] correspond to sets of vertices .
let @xmath306 be a set of graphs with uniformly regular tails .
for every @xmath300 there is an @xmath301 such that for each @xmath1619 we can find a set @xmath1620 corresponding to a set of vertices for @xmath72 such that @xmath1621 and @xmath1622 . [ prop26 ] since @xmath306 has uniformly regular tails we can find an @xmath301 such that for each @xmath1619 there is a set @xmath1623 ( not necessarily corresponding to a set of vertices for @xmath72 ) such that @xmath1624 and @xmath1625 . recall that each vertex @xmath1626 corresponds to an interval @xmath1627 for the stretched canonical graphon @xmath237 , such that @xmath1628 is proportional to the weight of the vertex . given a set @xmath1623 as above ,
define @xmath1629 the lemma now follows by observing that @xmath1630 and @xmath1631 define @xmath1632 ^ 2 \}$ ] .
if @xmath74 is sparse , then @xmath1633 . by lemma [ prop26 ] ,
if @xmath74 has uniformly regular tails there exists an @xmath1634 such that if we order the vertices of @xmath220 appropriately when defining the canonical graphon @xmath1635 of @xmath220 , then @xmath1636 ^ 2}\|_1>1/2 $ ] for all @xmath290 .
the graphons @xmath1637 and @xmath1638 are related by @xmath1639 for some @xmath1640 ( with @xmath1641 if @xmath220 has no isolated vertices ; if @xmath220 has isolated vertices corresponding to the end of the interval @xmath10 $ ] for the canonical graphon @xmath1635 we will have @xmath1642 ) . if @xmath1643 and @xmath1636 ^ 2}\|_1>1/2 $ ] for all @xmath290 , then @xmath1644 ^ 2}\|_1>1/2 \qquad\text{and}\qquad \lim_{n{\rightarrow}\infty } a_n=0,\qquad \text{where } a_n:= \min\big(m ' { \widetilde}m_n^{-1},1\big ) .
\label{eq36}\ ] ] the proof of ( i ) is complete if we can prove that implies that @xmath74 is not uniformly upper regular . assume the opposite , and let @xmath1645 and @xmath1646 be as in the definition of uniform upper regularity .
let @xmath1647 be a partition of @xmath0 such that one of the parts is @xmath1648 $ ] , where @xmath1649 is chosen as small as possible such that @xmath1648 $ ] corresponds to an integer number of vertices of @xmath220 for the canonical graphon .
then @xmath1650 since @xmath1651 and @xmath1652 .
by the first part of it follows that @xmath1653 on @xmath1648 ^ 2 $ ] for all sufficiently large @xmath16 ; hence for all sufficiently large @xmath16 , @xmath1654 ^ 2}\|_1 = \|w^{g_n , r}{\mathbf{1}}_{[0,a'_n]^2}\|_1 > 1/2.\ ] ] we have obtained a contradiction to the assumption of uniform upper regularity , and thus the proof of ( i ) is complete . defining @xmath1655 ( recall the definition of @xmath1656 in the beginning of section [ sec2 ] ) , we have @xmath1657 and @xmath1658 . if @xmath1659 is dense and has convergent edge density the following limit exists and is positive : @xmath1660 .
it follows by lemma [ prop16 ] ( resp.lemma [ prop15 ] ) that @xmath74 is a cauchy sequence for @xmath269 ( resp .
@xmath339 ) iff @xmath1661 is a cauchy sequence for @xmath105 , since for any @xmath1662 , @xmath1663 as @xmath1664 , and a similar estimate holds with @xmath339 instead of @xmath269 .
this completes the proof of the first assertion of ( ii ) . to prove the second assertion of ( ii )
consider the following sequence of dense graphs @xmath74 , which is a cauchy sequence for @xmath269 but not for @xmath339 .
for odd @xmath16 let @xmath220 be a complete simple graph on @xmath16 vertices , and for even @xmath16 let @xmath220 be the union of a complete graph on @xmath1665 vertices and @xmath1665 isolated vertices .
this sequence converges to @xmath296 ^ 2},{\mathbb{r}}_+)$ ] for @xmath269 , but does not converge for @xmath339 .
conversely , the following sequence of dense graphs @xmath74 is a cauchy sequence for @xmath339 but not for @xmath269 . for odd @xmath16
let @xmath220 be a complete graph on @xmath16 vertices , and for even @xmath16 let @xmath220 be an erds - rnyi graph with edge probability @xmath322 .
this sequence converges to @xmath59 for @xmath339 , but does not converge for @xmath269 .
we will assume throughout the proof that the graphs have no isolated vertices , since the case of @xmath1666 vertices clearly follows from this case .
we assume the average degree of @xmath74 is bounded above by @xmath1667 , and want to obtain a contradiction .
when defining the canonical stretched graphon @xmath1436 of @xmath220 , each vertex of @xmath220 corresponds to an interval of length @xmath1668 .
since @xmath1669 by assumption , the vertices of @xmath220 correspond to an interval of length @xmath1670 , which is too stretched out to be compatible with uniformly regular tails .
explicitly , given that @xmath220 has no isolated vertices it follows that for any @xmath301 and any borel set @xmath1671 satisfying @xmath1672 , @xmath1673 by the assumption that @xmath388 , the right side of this equation is greater than @xmath1674 for all sufficiently large @xmath290 .
since @xmath301 was arbitrary , this is not compatible with @xmath220 having uniformly regular tails , which together with theorem [ prop1 ] gives a contradiction .
the main goal of this section is to prove theorem [ prop5 ] .
let @xmath297 be a sequence of simple graphs with uniformly regular tails , such that @xmath1675 for all @xmath290 and @xmath1676 .
fix @xmath1667 , and for each @xmath290 let @xmath1393 be an induced subgraph of @xmath292 where all or some of the vertices of degree at most @xmath1677 are removed
. then @xmath1678 .
[ prop12 ] we wish to prove that @xmath1679 , where @xmath1680 . by taking a subsequence
we may assume @xmath1681 .
we will carry out the proof by contradiction , and assume @xmath300 . by definition of @xmath1393
there are at least latexmath:[${\varepsilon}_n at least one endpoint of degree at most @xmath1677 .
hence there are at least @xmath1683 vertices with degree between @xmath70 and @xmath1677 . in the canonical
stretched graphon of @xmath292 , each vertex corresponds to an interval of length @xmath1684 .
hence the total length of the intervals corresponding to vertices of degree between @xmath70 and @xmath1677 is at least @xmath1685 , which tends to infinity as @xmath222 .
it follows that for each @xmath301 and any sets @xmath1686 of measure at most @xmath1599 , @xmath1687 which is at least @xmath1688 when @xmath16 is sufficiently large .
thus @xmath1689 does not have uniformly regular tails , and we have obtained the desired contradiction
. we will now prove theorem [ prop5 ] .
note that we use ( * ? ? ?
* theorem 9.25 ) for part of the argument , a result which is also used in @xcite , but that we use it to prove theorem [ prop5 ] , which characterizes exchangeable random graphs that have uniformly regular tails , while in @xcite it is used to characterize exchangeable random graphs that have finitely many edges for each finite time , but which do not necessarily have uniformly regular tails ( a notion not considered in @xcite ) .
first assume @xmath25 is a graphon process generated by @xmath1690 with isolated vertices , where @xmath436 is a random variable .
we want to prove that @xmath25 has uniformly regular tails , and that the measure @xmath414 is exchangeable .
regularity of tails is immediate from theorems [ prop4](i ) and [ prop1 ] .
exchangeability is immediate by observing that the poisson random measure @xmath390 on @xmath46 defined in the beginning of section [ sec : w - random - results ] is identical in law to @xmath1691 for any measure - preserving transformation @xmath1692 ( in particular , for the case when @xmath478 corresponds to a permutation of intervals ) . to prove the second part of the theorem
assume that @xmath414 is jointly exchangeable and that @xmath25 has uniformly regular tails .
by joint exchangeability of @xmath414 it follows by the representation theorem for jointly exchangeable random measures on @xmath52 ( * ? ? ? * theorem 9.24 ) that a.s .
@xmath1693 for some measurable functions @xmath1694 on @xmath1162 , @xmath1695 on @xmath1696 , and @xmath1697 on @xmath52 , a set of independent uniform random variables @xmath1698 with values in @xmath10 $ ] , independent unit rate poisson processes @xmath1699 and @xmath1700 on @xmath52 for @xmath944 and @xmath1701 on @xmath1696 , an independent set of random variables @xmath1702 , and @xmath136 ( resp .
@xmath1703 ) denoting lebesgue measure on @xmath0 ( resp . the diagonal @xmath1704 ) . by the definition of @xmath414 as a sum of point masses , all the terms in involving lebesgue measure must be zero ; i.e. , except on an event of probability zero , @xmath1705 and @xmath1706 for all @xmath687 . recall that by each vertex can be uniquely identified with the time @xmath38 when it appeared in the graph , and that each point mass @xmath1707 with @xmath1708 represents an edge between the two vertices associated with @xmath50 and @xmath1709 . almost surely no two of the random variables @xmath1710 for @xmath1711 have the same value , and hence the functions @xmath1712 take values in @xmath1363 almost everywhere .
furthermore , since the graphs @xmath26 are undirected , we have @xmath1713 and @xmath1714 , and @xmath1190 is symmetric in its second and third input argument .
first we will argue that the subgraphs @xmath1242 of @xmath26 corresponding to the terms @xmath1715 have the law of a graphon process with isolated vertices generated by some ( possibly random ) graphon @xmath158 .
condition on the realization of @xmath436 , and define the function @xmath1716 $ ] by @xmath1717 it follows that , conditioned on @xmath436 such that @xmath1718 , @xmath1719 has the law of a graphon process generated by @xmath437 . to conclude we need to prove that @xmath1720 almost surely , which will be done in the next two paragraphs .
first we will argue that @xmath1719 has uniformly regular tails .
since no two of the random variables @xmath1710 have the same value for @xmath1711 , each point mass @xmath1721 or @xmath1722 of @xmath414 corresponds to an isolated edge , i.e. , an edge between two vertices each of degree one , and each point mass @xmath1723 or @xmath1724 of @xmath414 corresponds to an edge between the vertex @xmath1725 in @xmath1242 and a vertex of degree one ; i.e. , @xmath1726 corresponds to a star centered at the vertex associated with @xmath1725 . note that @xmath1242 and @xmath26 satisfy the conditions of lemma [ prop12 ] with @xmath1727 .
hence @xmath1728 , and since @xmath26 has uniformly regular tails this implies that @xmath1242 must also have uniformly regular tails .
we assume that @xmath1729 is not almost surely integrable , and will derive a contradiction .
we condition on @xmath436 such that @xmath1730 , and to simplify notation we will write @xmath14 instead of @xmath1729 . let @xmath1731 ( resp .
@xmath1732 ) be the induced subgraph of @xmath1242 consisting of the vertices for which the feature @xmath40 satisfies @xmath1733 ( resp .
@xmath1734 ) .
let @xmath1735 and @xmath1736 denote the corresponding graphons ( we will see shortly that they are integrable ) , i.e. , @xmath1737 and @xmath1738 . since @xmath1739 a.s . for each @xmath38
the measure @xmath414 is locally finite a.s . hence ( * ? ? ?
* theorem 9.25 ) ( iii ) and ( iv ) , respectively , imply that @xmath1740 and @xmath1741 . by applying proposition [ prop2 ] this implies further that @xmath1742 @xmath1743 and @xmath1744 it follows that if @xmath1745 is the set of edges having one endpoint in @xmath1746 and one endpoint in @xmath1747 , we have @xmath1748 for the stretched canonical graphon @xmath1749 the edges @xmath1750 correspond to @xmath1751 , where @xmath1752 corresponds to @xmath1753 .
since @xmath1754 , we have @xmath1755 . by and @xmath1756
, we have @xmath1757 .
since @xmath1758 and @xmath1759 takes values in @xmath10 $ ] , we have @xmath1760 for all sets
@xmath1761 of bounded measure .
we have obtained a contradiction to the hypothesis of uniform regularity of tails , since @xmath1762 to complete the proof that @xmath25 has the law of a graphon process with isolated vertices , we need to argue that a.s .
@xmath1763 let @xmath1764 denote the number of edges associated with terms of @xmath414 of the form @xmath1765 , and let @xmath1766 denote the number of edges associated with terms of @xmath414 of the form @xmath1767 .
since @xmath1242 and @xmath26 satisfy the conditions of lemma [ prop12 ] with @xmath1727 , and since lemma [ prop2 ] implies that a.s.-@xmath1768 , it follows that a.s.@xmath1769 we will prove by contradiction , and will consider each term separately .
first assume @xmath1770 with positive probability .
conditioned on a realization of @xmath436 such that @xmath1771 , the random variable @xmath1772 is a poisson random variable with expectation @xmath1773 .
hence @xmath1774 , which is a contradiction to .
it follows that @xmath1775 a.s . , and thus @xmath1776 for all @xmath89 a.s .
now assume @xmath1777 with positive probability .
then there exists @xmath300 such that with positive probability there is a set @xmath1778 ( depending on @xmath436 ) satisfying @xmath1779 , and such that for all @xmath1780 it holds that @xmath1781 , where @xmath1782 .
consider the poisson point process @xmath1699 corresponding to the graphon process @xmath26 with isolated vertices .
the number of points @xmath1783\times i$ ] evolves as a function of @xmath50 as a poisson process with rate @xmath300 ; hence the number of such points divided by @xmath50 converges to @xmath710 a.s . for any given pair @xmath1783\times i$ ]
the number of points @xmath1784\times i_{x_j}$ ] for the poisson point process @xmath1785 has the law of a poisson random variable with intensity greater than @xmath1786 .
hence , @xmath1787 this contradicts , and thus completes our proof that @xmath25 has the law of a graphon process with isolated vertices . in our proof
above we observed that the assumption of exchangeability alone is not sufficient to prove that @xmath25 has the law of a graphon process with isolated vertices .
more precisely , without this assumption we might have @xmath1788 and the measure might also consist of the terms containing @xmath1789 and @xmath1790 .
we observed in the proof that the terms containing @xmath1790 correspond to isolated edges , and that the terms containing @xmath1789 correspond to `` stars '' centered at a vertex in the graphon process .
it is outside the scope of this paper to do any further analysis of these more general exchangeable graphs .
in this section we will prove proposition [ prop3 ] .
the following lemma will imply part ( ii ) of the proposition .
let @xmath15 be a bounded , non - negative graphon , and assume that @xmath1791 for a star @xmath1792 with @xmath1233 leaves .
then @xmath495 for all simple , connected graphs @xmath475 of maximal degree at most @xmath1233 .
[ lem : h - bd ] we first note that if @xmath1793 for a star with @xmath1233 leaves , then the same holds for all stars with at most @xmath1233 leaves , since we know that @xmath1794 by our definition of a graphon . also , using that @xmath14 is bounded , we assume without loss of generality that @xmath475 is a tree @xmath1795 of maximal degree @xmath1796 .
designate one of the vertices , @xmath1797 , as the root of the tree , and choose a vertex @xmath1798 such that no other vertex is further from the root .
if @xmath1798 has distance @xmath70 from @xmath1797 , then @xmath1795 is a star and there is nothing to prove .
let @xmath1799 be @xmath1798 s grandparent , let @xmath376 be its parent , and let @xmath1800 for @xmath1801 be its siblings .
note that by our assumption on @xmath1798 , all the siblings @xmath1802 are leaves .
furthermore , if their grandparent @xmath1799 is the root and the root has no other children , then @xmath1795 is again a star , so we can rule that out as well .
if we remove the edge @xmath1803 from @xmath1795 , we obtain two disjoint trees @xmath1804 and @xmath1805 , and as just argued , the one containing @xmath1799 is a tree with at least @xmath1806 vertices and maximal degree at most @xmath1807 , while the second one is a star , again of maximal degree at most @xmath1807 . because @xmath1808 , the lemma now follows by induction .
we will start by proving ( i ) .
fix some simple connected graph @xmath475 with @xmath1233 vertices . by proposition [ prop2 ] applied with @xmath475 and the simple connected graph on two vertices , respectively , @xmath1809 and
hence @xmath1810 proving ( i ) .
\(ii ) since @xmath1811 if @xmath475 is a star with @xmath1233 leaves , we can use lemma [ lem : h - bd ] to conclude that @xmath501 for every simple connected graph @xmath475 with at least two vertices . express @xmath1812 as @xmath1813 , where we sum over all equivalence relations @xmath1814 on @xmath1235 . by proposition [ prop2 ] applied with @xmath1815 , we have @xmath1816 unless @xmath1814 is the equivalence relation for which the number of equivalence classes equals @xmath1817 .
hence the estimate holds with @xmath1818 in place of @xmath1819 , which completes the proof of ( ii ) .
next we will prove ( iii ) .
let @xmath475 be a simple connected graph with at least three vertices , and assume @xmath1667 is such that the degree of the vertices of @xmath220 is bounded by @xmath1677 .
we may assume @xmath220 has no isolated vertices , since @xmath1820 is invariant under adding or deleting isolated vertices .
under the assumption of no isolated vertices , we have latexmath:[$|e(g_n)|\geq latexmath:[${\operatorname}{hom}(f , g_n)\leq latexmath:[$h(f , g_n ) \leq follows .
now we will prove ( iv ) .
we first construct an example of a sequence of graphs @xmath74 which converges for the stretched cut metric @xmath269 , but which is not left convergent .
let @xmath297 be a sequence of simple dense graphs with @xmath1822 that is convergent in the @xmath105 metric , and hence also in the @xmath269 metric .
define @xmath1823 for even @xmath16 , and for odd @xmath16 let @xmath220 be the union of @xmath292 and @xmath1824 vertices of degree one , which are all connected to the same uniformly random vertex of @xmath292 .
then @xmath74 converges for @xmath269 with the same limit as @xmath297 , since @xmath292 is an induced subgraph of @xmath220 and @xmath1825 . on the other hand , @xmath74 is not left convergent , since if @xmath475 is the simple connected graph with three vertices and two edges , then @xmath1826 for odd @xmath16 and
hence @xmath1827 along sequences of odd @xmath16 , while @xmath1820 converges to a finite number by the fact that dense graph sequences which are convergent in the cut metric are left convergent .
finally we will provide a counterexample in the reverse direction ; i.e. , we will construct a sequence of graphs @xmath74 which is left convergent , but does not converge for the stretched cut metric .
let @xmath297 be left convergent and satisfy @xmath1676 , and let @xmath220 be the union of @xmath292 and @xmath1828 isolated edges .
then @xmath74 is left convergent , since @xmath1829 when @xmath475 is the simple connected graph on two vertices , and @xmath1830 when @xmath475 has at least three vertices . on the other hand , @xmath74 is not convergent for @xmath269 , since it does not have uniformly regular tails .
we thank svante janson for his careful reading of an earlier version of the paper and for his numerous suggestions , and we thank edoardo airoldi for helpful discussions . c. borgs , j. t. chayes , l. lovsz , v. t. ss , and k. vesztergombi , _ convergent sequences of dense graphs i : subgraph frequencies , metric properties and testing _ , adv .
* 219 * ( 2008 ) , no . 6 , 18011851 . s. janson , _
graphons , cut norm and distance , couplings and rearrangements _ , new york journal of mathematics .
nyjm monographs , vol . 4 , state university of new york , university at albany , albany , ny , 2013 . | in a recent paper , caron and fox suggest a probabilistic model for sparse graphs which are exchangeable when associating each vertex with a time parameter in @xmath0 . here
we show that by generalizing the classical definition of graphons as functions over probability spaces to functions over @xmath1-finite measure spaces , we can model a large family of exchangeable graphs , including the caron - fox graphs and the traditional exchangeable dense graphs as special cases .
explicitly , modelling the underlying space of features by a @xmath1-finite measure space @xmath2 and the connection probabilities by an integrable function @xmath3 $ ] , we construct a random family @xmath4 of growing graphs such that the vertices of @xmath5 are given by a poisson point process on @xmath6 with intensity @xmath7 , with two points @xmath8 of the point process connected with probability @xmath9 .
we call such a random family a _ graphon process_. we prove that a graphon process has convergent subgraph frequencies ( with possibly infinite limits ) and that , in the natural extension of the cut metric to our setting , the sequence converges to the generating graphon .
we also show that the underlying graphon is identifiable only as an equivalence class over graphons with cut distance zero .
more generally , we study metric convergence for arbitrary ( not necessarily random ) sequences of graphs , and show that a sequence of graphs has a convergent subsequence if and only if it has a subsequence satisfying a property we call _ uniform regularity of tails_. finally , we prove that every graphon is equivalent to a graphon on @xmath0 equipped with lebesgue measure . |
scalar multiplication on elliptic curves ( or jacobians of genus 2 curves ) is a key operation in many modern asymmetric cryptographic primitives .
the classic scenario is as follows : let @xmath1 be a cyclic subgroup of order @xmath2 , where @xmath3 is an elliptic curve or an abelian surface over a finite field @xmath4 . given an integer @xmath5 ( typically on the order of @xmath2 ) and a point @xmath6 in @xmath7 , our goal is to compute @xmath8p : = \underbrace{p + p + \cdots + p}_{m \text { times } } \ ] ] as quickly as possible . since elliptic curve scalar multiplication is analogous to exponentiation in finite fields , many algorithms originally developed with the multiplicative groups of finite fields ( or general abelian groups ) in mind transfer directly to scalar multiplication : square - and - multiply loops in finite fields
become double - and - add loops on elliptic curves , for example .
however , the geometry of elliptic curves can offer us new algorithms with no true finite field analogues .
a spectacular ( and easy ) example of this phenomemon is scalar multiplication with endomorphism decompositions , originally proposed by gallant , lambert , and vanstone @xcite .
we present a general version of their idea below that is flexible enough to accommodate higher - dimensional and higher - genus constructions .
let @xmath3 , @xmath7 , and @xmath2 be as above , and let @xmath9 be @xmath4-endomorphisms of @xmath3 .
we lose nothing by supposing @xmath10 .
in contrast to @xcite , we do _ not _ suppose that the @xmath11 form a linearly independent set .
suppose that @xmath12 for @xmath13 ( this is the typical situation in cryptographic applications , where @xmath2 is so close to @xmath14 that there is no room for the image of @xmath7 to be anything but @xmath7 itself ) ; then each @xmath11 restricts to an endomorphism of @xmath7 . but @xmath7 is a cyclic group , isomorphic to @xmath15 , and as such each of its endomorphisms is multiplication by some integer ( defined modulo @xmath2 ) .
in particular , each endomorphism @xmath11 acts on @xmath7 as multiplication by an integer _ eigenvalue _
@xmath16 , such that @xmath17_{\mathcal{g}}\ .\ ] ] let @xmath18 be the @xmath19-module homomorphism @xmath20 an @xmath21-dimensional of a decomposition has no relation to the dimension of @xmath3 , or to the @xmath19-rank of the endomorphism ring .
typical values for @xmath21 are @xmath22 , corresponding to classical scalar multiplication ; @xmath23 , as in in gallant lambert vanstone @xcite and galbraith lin
scott @xcite multiplication ; and @xmath24 , as proposed by longa and sica @xcite and guillevic and ionica @xcite .
a technique with @xmath25 was proposed by zhou , hu , xu , and song @xcite , but this is essentially longa sica with @xmath26 .
bos , costello , hisil , and lauter have implemented a genus 2 scalar multiplication with @xmath27 , but this seems to be the upper limit of practicality for these techniques @xcite . ] _ decomposition _ of a scalar @xmath5 is any element of @xmath28 .
returning to the scalar multiplication problem : we can compute @xmath29p$ ] for any @xmath6 in @xmath7 by using a multiexponentiation algorithm on the points @xmath30 to compute @xmath8p = [ a_1]\phi_1(p ) \oplus \cdots \oplus [ a_r]\phi_r(p ) \qquad \text{for any } \quad ( a_1,\ldots , a_r ) \in d^{-1}(m ) \ .\ ] ] the literature on exponentation and multiexponentiation algorithms is vast , and we will not attempt to summarize it here ( but for an introduction to general exponentation and multiexponentation algorithms , we recommend ( * ? ? ?
* , 11.2 ) and ( * ? ? ?
* chapter 9 ) ) . for the purposes of this article
, it suffices to note that for the scalar decomposition technique to offer an advantage over simply computing @xmath29p$ ] as a conventional exponentiation , the endomorphisms must be _ efficient _ : : that is , any @xmath31 must be computable for the cost of a few group operations , and the decomposition must be _ short _ : : that is , @xmath32 should be significantly smaller than @xmath33 , which is typically on the order of @xmath2 . in this article , we suppose that we are given a fixed set of efficient @xmath11 , and concentrate on the problem of computing short scalar decompositions .
first , consider the lattice of decompositions of @xmath34 : @xmath35 the set of decompositions of any @xmath5 in @xmath15 is then the lattice coset @xmath36 to find a short decomposition of @xmath5 , we can subtract a nearby vector in @xmath37 from @xmath38 . the reference technique for finding such a vector in @xmath37
is babai rounding @xcite , which works as follows : if @xmath39 is a basis for @xmath37 , then we let @xmath40 be the ( unique ) solution in @xmath41 to the linear system @xmath42 and set @xmath43 then @xmath44 is an @xmath21-dimensional decomposition of @xmath5 . since @xmath45 and @xmath46 for any @xmath47 in @xmath0 , we have @xmath48 it is clear , therefore , that finding short decompositions depends on finding a short basis for @xmath37 .
note that @xmath37 depends only on the @xmath11 , and not on the eventual scalars @xmath5 or points @xmath6 to be multiplied ; as such , the short basis can ( and should ) be precomputed . assuming that the eigenvalues have pairwise differences of absolute value at least @xmath49 , there exists a basis with @xmath50 in @xmath51 , which will yield scalar decompositions of bitlength around @xmath52 . in most of the scalar decomposition literature ,
a short basis of @xmath37 is produced by starting with a long basis typically the basis @xmath53 before applying a lattice reduction algorithm to produce a short basis .
for @xmath23 , gallant , lambert , and vanstone used the euclidean algorithm , which is equivalent to the usual gauss lattice reduction algorithm ( though kaib s algorithm for the infinity norm @xcite may give marginally better results ) .
in higher dimensions , we would typically use a fast lll variant ( such as ` fplll ` @xcite)though longa and sica @xcite went so far as to propose a new 4-dimensional lattice basis reduction algorithm for their glv+gls construction on elliptic curves .
our contention in this article is that in most cryptographic situations , no lattice basis reduction is required to find a short basis of @xmath37 : one can simply write down vectors of length at most @xmath54 from scratch .
the information that allows us to do so is typically a by - product of the group order computation ( or of the cm method ) . for the abelian varieties most useful in cryptography ,
these vectors either form a basis for @xmath37 , or can be easily modified to do so .
galbraith , lin , and scott @xcite and the author @xcite have already constructed families of endomorphisms equipped with a convenient ready - made basis ; in this work , we generalize these ready - made bases to all of the other known efficient endomorphism constructions for elliptic curves and to real multiplication techniques for genus 2 jacobians . in this way , we construct explicit short bases for the galbraith
scott ( gls ) , gallant lambert vanstone ( glv ) , guillevic ionica @xcite , longa
sica , and @xmath0-curve reduction techniques , as well as for the kohel
smith @xcite and takashima @xcite methods for genus 2 jacobians .
we do not pretend that this is a significant optimization for scalar decomposition methods : the construction of a short basis is essentially a one - shot precomputation , and existing lattice basis reduction methods are certainly fast enough on the relevant input sizes .
however , the construction of these `` instant '' bases turns out to be an illuminating exercise : short bases can be read off from they are simple endomorphism ring relations that are , in practice , known in advance .
the bottom line is that it is always more convenient to not compute something than it is to compute it .
we recall some elementary facts from the theory of quadratic fields .
further details and proofs can be found in almost any basic algebraic number theory text ( we recommend @xcite ) .
let @xmath55 be a quadratic field , real or imaginary , with maximal order @xmath56 and discriminant @xmath57 .
if @xmath58 is an element of @xmath56 then we write @xmath59 for its trace , @xmath60 for its norm .
if @xmath58 is not in @xmath19 , then it generates an order @xmath61 $ ] in @xmath56 ; we write @xmath62 for the discriminant of @xmath61 $ ] , and @xmath63 for the minimal polynomial of @xmath58 .
the discriminants of @xmath56 and @xmath61 $ ] are related by @xmath64 for some positive integer @xmath65 , the conductor of @xmath61 $ ] in @xmath56 .
the set of orders in @xmath55 form a lattice ( in the combinatorial sense ) , indexed by the conductor : @xmath61 \subset { \mathbb{z}}[\xi']$ ] if and only if @xmath66 .
if @xmath61 \subset { \mathbb{z}}[\xi']$ ] are orders in @xmath55 , then necessarily @xmath67 for some integers @xmath68 and @xmath69 .
it follows that @xmath70 note that @xmath69 is , up to sign , the relative conductor of @xmath61 $ ] in @xmath71 $ ] .
multiplying eq . through by @xmath72 , which is also @xmath73
, we obtain a second relation @xmath74 the following lemma turns the relations between endomorphisms of eqs . and into relations between eigenvalues , which we will use later to produce short lattice vectors .
[ lemma : relations ] let @xmath58 and @xmath75 be endomorphisms of an abelian variety @xmath76 such that @xmath61 $ ] and @xmath71 $ ] are quadratic rings and @xmath61 \subseteq { \mathbb{z}}[\xi']$ ] , so @xmath77 for some integers @xmath68 and @xmath69 .
let @xmath78 be a cyclic subgroup of order @xmath2 such that @xmath79 and @xmath80 , and let @xmath81 and @xmath82 be the eigenvalues in @xmath15 of @xmath58 and @xmath75 on @xmath7 , respectively .
then @xmath83 this follows immediately by mapping the relations in eqs . and through the homomorphism @xmath71\to{\mathrm{end}}({\mathcal{g } } ) \cong { \mathbb{z}}/n{\mathbb{z}}$ ] sending @xmath75 to @xmath84 ( and @xmath58 to @xmath85 ) .
let @xmath86 be an ordinary elliptic curve . if @xmath87 is the @xmath88-power frobenius endomorphism on @xmath89 then @xmath90 where @xmath91 [ th : ec-2d ] let @xmath92 be a non - integer endomorphism of @xmath89 such that @xmath93 \subset { \mathbb{z}}[\phi]$ ] , so @xmath94 for some integers @xmath69 and @xmath68 .
suppose that we are in the situation of [ sec : general ] with @xmath95 and @xmath96 .
the vectors @xmath97 generate a sublattice of @xmath37 of determinant @xmath98 . if @xmath99 , then @xmath100 .
the frobenius endomorphism @xmath87 fixes the points in @xmath101 , so @xmath102 and the eigenvalue of @xmath87 on @xmath7 is @xmath103 .
applying lemma [ lemma : relations ] with @xmath104 and @xmath105 , we obtain relations @xmath106 the first implies that @xmath107 is in @xmath37 , the second that @xmath108 is in @xmath37 .
equations imply that @xmath109 which is @xmath110 .
theorem [ th : ec-2d ] constructs two basis vectors , but it makes no claim about their length
. we can give some almost - trivial bounds on the size of @xmath68 and @xmath69 , using @xmath111 , @xmath112 , and the triangle inequality : @xmath113 but proving general bounds that apply for arbitrary endomorphisms is probably the wrong approach when all of the endomorphisms used in practical scalar decompositions are special ( and deliberately non - general ) constructions .
consider the ciet
quisquater bounds of ( * ? ? ?
* theorem 1 ) for the case @xmath23 of 1 : for every integer @xmath5 , there exists a decomposition @xmath114 of @xmath5 such that @xmath115 , with @xmath116 . in terms of bitlength , @xmath117 in this theorem , @xmath118 and @xmath119 ( and @xmath120 ) are implicitly treated as constants that is , independent of @xmath2 and @xmath88 . while this is appropriate for glv curves , @xmath118 and @xmath119 are not `` constant '' when @xmath92 is inseparable ( notably , @xmath119 is divisible by @xmath121 ) . in this case , the bound above is spectacularly loose : remark [ rem : gls - bounds ] below gives a detailed example of this in the context of gls endomorphisms .
let @xmath122 be the cofactor such that @xmath123 . by the pohlig
silver reduction @xcite , the cryptographic strength of @xmath89 depends entirely on the size of the prime @xmath2 , so we should choose @xmath89 with @xmath2 as large as possible . while @xmath124 is ideal , allowing @xmath125 or @xmath126 permits faster curve arithmetic via transformations to montgomery @xcite or twisted edwards @xcite models , for example .
( in the pairing - based context @xmath122 may be somewhat larger . )
the lattice @xmath37 has determinant @xmath2 , but the vectors @xmath127 constructed by theorem [ th : ec-2d ] are a basis for a ( sub)lattice of determinant @xmath128 .
if @xmath129 , then while our sublattice will still give short decompositions , it is suboptimal .
if @xmath130 , for example , then our basis may be one bit too long which generally means one double too many when the resulting decompositions are used in a multiexponentiation algorithm .
if for some reason @xmath131 , then we want to be able to derive a basis of the full lattice @xmath37 from @xmath127 .
first , note that @xmath132 because the relations @xmath133 and @xmath134 still hold when we divide through by @xmath135 .
this new sublattice has index @xmath136 in @xmath37 .
note that if @xmath137 , then @xmath138 \subset { \mathcal{e}}({\mathbb{f}}_{q})$ ] , because @xmath139 , so @xmath140 factors through @xmath141 $ ] . more generally , suppose @xmath142 is a prime dividing @xmath122 : then there exists a sublattice @xmath143 such that @xmath144 = \ell \ .\ ] ] looking at the components of @xmath107 and @xmath108 , we see that if @xmath142 divides @xmath69 , then it must also divide @xmath145 ( otherwise @xmath143 can not exist ) , \subseteq { \mathcal{e}}({\mathbb{f}}_{q})$ ] then @xmath142 must divide the conductor of @xmath93 $ ] in the endomorphism ring of @xmath89 ( see also @xcite ) . in this case @xmath142
divides both @xmath140 and @xmath146 , so it must also divide @xmath145 . removing
the factor of @xmath147 from the index of our sublattice therefore corresponds to removing the contribution of the full @xmath142-torsion to our endomorphism relations . ] and we can replace each @xmath148 with @xmath149 as above to produce a sublattice of index @xmath150 in @xmath37 .
suppose now that @xmath142 does not divide @xmath69 .
if @xmath142 divides @xmath145 and @xmath151 , then @xmath152 is in @xmath37 , so we can take @xmath153 . otherwise , @xmath154 is in @xmath37 for precisely one @xmath155 : that is , @xmath156
. we can therefore take @xmath157 .
iterating this process factor by factor of @xmath122 , we can gradually shrink the vectors produced by theorem [ th : ec-2d ] to derive a true basis of @xmath37 .
but as we remarked above , in conventional discrete - log based cryptography the most important cases are @xmath158 , and @xmath126 , and then there is almost nothing to be done . to handle the cases
where @xmath159 or @xmath160 , for example , the following simple procedure produces a basis for @xmath37 : * if @xmath69 is even , then @xmath161 . *
if @xmath69 and @xmath68 are odd and @xmath151 is even , then @xmath162 .
* otherwise , @xmath163 .
let @xmath164 be an elliptic curve over @xmath165 with complex multiplication by @xmath166 $ ] ; that is , with an explicit endomorphism @xmath167 such that @xmath168 \cong { \mathbb{z}}[\sqrt{\delta}]$ ] .
let @xmath86 be the ( good ) reduction modulo @xmath121 of @xmath89 , ( and suppose that @xmath89 is not supersingular ) ; by definition , @xmath89 comes equipped with an explicit separable endomorphism @xmath92 such that @xmath169 \cong { \mathbb{z}}[\sqrt{\delta}]$ ] .
if @xmath170 and @xmath164 were was chosen in such a way that @xmath92 has very low degree , then @xmath92 can be efficiently computable , and hence useful for scalar decompositions .
suppose , therefore , that @xmath92 has very low degree for efficiency reasons
. then @xmath171 must be very small ; and since @xmath172 must be negative , @xmath173 ( and hence @xmath174 ) must also be very small .
in particular , @xmath175 and @xmath176 . in practice ,
@xmath169 $ ] is either the maximal order in @xmath177 , or ( exceptionally ) an order of index two in the maximal order .
it is therefore reasonable to assume @xmath93 $ ] is contained in @xmath169 $ ] , so @xmath178 this allows us to write down a basis for ( a sublattice of ) the glv lattice using theorem [ th : ec-2d ] .
it is important to note that in practice , @xmath68 and @xmath69 are already known from the determination of the curve order , precisely because @xmath94 .
indeed , if we want to compute the order of an elliptic curve known to have complex multiplication by a cm order ( such as @xmath169 $ ] , in this case ) , then we would typically use the algorithm described in @xcite , which computes the trace @xmath179 of frobenius _ by computing @xmath68 and @xmath69_. this approach uses cornacchia s algorithm to compute a generator of a principal ideal in @xmath169 $ ] of norm @xmath88 , before taking its trace to compute @xmath98 ; but this generator is none other than @xmath94 . alternatively , @xmath89 could be constructed using the cm method starting from the small discriminant @xmath172 . in this case , @xmath68 and @xmath69 are explicitly constructed so that @xmath180 will have norm @xmath88 . in any case ,
if we had somehow mislaid the values of @xmath68 and @xmath69 , then we could recover them by factoring the ideal @xmath181 in @xmath169 $ ] using ( for example ) cornacchia s algorithm , which amounts to repeating the point counting algorithm described above . the element @xmath180 will be ( up to sign ) one of the two resulting generators of the factors of @xmath181 .
alternatively , we could use @xmath182 ; though inverting @xmath183 modulo @xmath2 is roughly equivalent to the use of the euclidean algorithm in the original glv method .
one important feature of the glv setting is that @xmath184 , @xmath185 , and @xmath172 are independent of @xmath88 and @xmath179 , so the bitlength of the basis produced by theorem [ th : ec-2d ] exceeds @xmath186 by no more than an explicit constant .
the following examples consider the new basis for the glv curves with endomorphisms of degree at most @xmath187 ( treated in detail elsewhere by gallant , lambert , and vanstone @xcite and longa and sica @xcite ) .
if @xmath188 , then for every @xmath189 in @xmath4 the curve @xmath190 has an @xmath4-endomorphism @xmath191 ( where @xmath192 ) , with @xmath193 ; so @xmath169 \cong { \mathbb{z}}[\sqrt{-1}]$ ] .
theorem [ th : ec-2d ] constructs the basis @xmath194 where @xmath195 .
this basis is not only short ( clearly @xmath196 ) and reduced , it is also orthogonal .
if @xmath197 , then for any @xmath189 in @xmath4 , the curve @xmath198 has an @xmath4-endomorphism @xmath199 ( where @xmath200 is a primitive third root of unity ) , with @xmath201 : that is , @xmath202 has explicit cm by @xmath203 $ ] .
looking at the basis produced by theorem [ th : ec-2d ] , we find @xmath204 where @xmath205 .
we note that in this case , two applications of the triangle inequality yields @xmath206 .
suppose @xmath207 is a square in @xmath4 .
the curve @xmath208 over @xmath4 has a degree-2 endomorphism @xmath92 with @xmath209 ( and @xmath210 ) ; that is , @xmath211 has explicit cm by @xmath212 $ ] .
theorem [ th : ec-2d ] yields vectors @xmath213 as before , @xmath214 .
suppose @xmath215 is a square in @xmath4 .
the curve @xmath216 over @xmath4 has a degree-2 endomorphism @xmath92 with @xmath217 ( and @xmath218 ) : that is , @xmath219 has explicit cm by @xmath220 $ ] .
theorem [ th : ec-2d ] yields vectors @xmath221 with @xmath222 .
suppose @xmath223 is a square in @xmath4 .
the curve @xmath224 over @xmath4 has a degree 3 endomorphism @xmath92 with @xmath225 : that is , @xmath226 has explicit cm by @xmath227 $ ] .
the kernel of @xmath92 is cut out by @xmath228 .
theorem [ th : ec-2d ] constructs a pair of vectors @xmath229 with @xmath230 .
suppose @xmath231 is a square in @xmath4 .
the curve @xmath232 over @xmath4 has an @xmath4-endomorphism @xmath92 of degree @xmath187 with minimal polynomial @xmath233 ( and kernel cut out by @xmath234 ) : that is , @xmath235 has explicit cm by @xmath236 $ ] .
theorem [ th : ec-2d ] yields vectors @xmath237 with @xmath214 .
let @xmath238 be an ordinary elliptic curve over @xmath239 , and let @xmath240 be the base extension of @xmath238 to @xmath241 .
the frobenius endomorphism @xmath242 of @xmath238 has characteristic polynomial @xmath243 with @xmath244 , while the ( @xmath245-power ) frobenius endomorphism @xmath87 of @xmath89 satisfies @xmath246 , so @xmath247 .
in particular , @xmath248 so @xmath249 can not have prime divisors larger than @xmath250 . now let @xmath251 be the quadratic twist of @xmath89 , with @xmath252 the twisting isomorphism .
the frobenius @xmath253 on @xmath251 satisfies @xmath254 , so @xmath255 note that @xmath256 ; unlike @xmath98 , this can take prime ( and near - prime ) values , so @xmath251 may be useful for discrete - logarithm - based cryptosystems .
the gls endomorphism on @xmath251 is @xmath257 .
it is defined over @xmath241 ; its minimal polynomial is @xmath258 if @xmath7 is a cyclic subgroup of @xmath259 of order @xmath2 such that @xmath260 , then the eigenvalue @xmath261 of @xmath262 on @xmath7 is a square root of @xmath263 modulo @xmath2 .
we have @xmath264 , so @xmath265 $ ] contains @xmath266 $ ]
. we can therefore apply theorem [ th : ec-2d ] to the inclusion @xmath266 \subseteq { \mathbb{z}}[\psi]$ ] in order to compute a short basis for ( a sublattice of ) @xmath267 .
looking at the discriminants , we see that @xmath266 $ ] has conductor @xmath268 in @xmath265 $ ] . indeed ,
@xmath269 so theorem [ th : ec-2d ] yields a basis @xmath270 this is precisely ( up to sign ) the basis of ( * ? ? ?
* lemma 3 ) ; it is not only short ( the largest coefficient is @xmath271 , so @xmath272 ) , it is also orthogonal . if @xmath259 does not have prime order , then we can easily shrink the basis to fit @xmath7 by following the procedure described in
[ sec : shrink ] .
[ rem : gls - bounds ] from a purely formal point of view , we could have treated this identically to the glv case , with @xmath262 in place of @xmath92 , but there are a number of important differences .
first of all , the ring @xmath265 $ ] has a much larger discriminant than any glv order : in general @xmath265 $ ] is far from being the maximal order of the endomorphism algebra .
second , the parameters @xmath273 , @xmath274 , and @xmath275 vary with @xmath121 and @xmath276 , so we can not treat the excess bitlength in the ciet sica quisquater bounds ( ineq . ) as a constant . indeed ,
if we simply plug the values @xmath277 into inequalities or , or even the `` optimal '' bound of ( * ? ?
* theorem 4 ) , then we obtain a rather pessimistic bitlength bound of around @xmath278 , which exceeds the true length of the basis by @xmath279 bits .
gls curves may be seen as a special case of a more general construction involving reductions of quadratic @xmath0-curves .
we give a very brief description of this construction here ( see @xcite for more details , and families of examples ) . in (
* proposition 2 ) , a basis is constructed in a seemingly ad - hoc way that yields half - length scalar decompositions ; we will see below that this basis also results from theorem [ th : ec-2d ] .
let @xmath55 be a quadratic field , @xmath280 the nontrivial automorphism of @xmath55 fixing @xmath0 .
let @xmath281 and @xmath282 be a pair of galois - conjugate curves over @xmath55 such that there exists an isogeny @xmath283 of small degree @xmath284 defined over @xmath285 . if @xmath121 is an inert prime in @xmath55 that is a prime of good reduction for @xmath286 ( and not dividing @xmath284 ) , then we can reduce @xmath287 modulo @xmath121 to obtain a @xmath284-isogeny @xmath288 of curves over @xmath241 . here
@xmath289 , the reduction of @xmath290 modulo @xmath121 , is the curve formed from @xmath89 by applying @xmath121-th powering to its coefficients ; so there also exists a @xmath121-th power frobenius isogeny @xmath291 .
composing @xmath92 with @xmath242 , we obtain an inseparable endomorphism @xmath292 of @xmath89 , of degree @xmath293 .
if @xmath284 is very small , then @xmath262 can be efficiently computable , since @xmath121-th powering in @xmath241 is essentially free .
( the gls construction is equivalent to the special case where @xmath92 is an isomorphism that is , @xmath294 . )
let @xmath295 ; that is , @xmath296 if @xmath297 is a nonsquare modulo @xmath121 , and @xmath263 if it is a square .
then according to ( * ? ? ?
* proposition 1 ) , the minimal polynomial of @xmath262 is @xmath298 and @xmath299 ( this determines @xmath21 up to sign ; exchanging @xmath21 with @xmath300 corresponds to exchanging @xmath92 with @xmath301 . ) squaring the endomorphism , we obtain @xmath302\pi$ ] , where @xmath87 is the usual @xmath245-power frobenius on @xmath89 .
in particular , @xmath93 $ ] is contained in @xmath265 $ ] ; we find that @xmath303 if @xmath304 is a cyclic subgroup of order @xmath2 such that @xmath305 , then the eigenvalue of @xmath262 on @xmath7 is a square root of @xmath306 modulo @xmath2 .
applying theorem [ th : ec-2d ] to the orders @xmath93\subseteq{\mathbb{z}}[\psi]$ ] , we obtain the basis @xmath307 up to sign , this is the basis appearing in the proof of ( * ? ? ?
* proposition 2 ) , where it is used to produce scalar decompositions having bitlength at most @xmath308 . while this basis generates a sublattice of determinant
@xmath249 , if @xmath309 then the basis may be easily shrunk to fit @xmath7 following the procedure outlined in
[ sec : shrink ] .
recently , longa and sica @xcite followed by guillevic and ionica @xcite have proposed using a pair of efficiently computable endomorphisms @xmath92 and @xmath262 to construct four - dimensional scalar decompositions on elliptic curves , corresponding to the situation of [ sec : general ] with @xmath310 .
( the longa sica technique supersedes the earlier 3-dimensional construction of zhou , hu , xu , and song @xcite with @xmath311 , which we will not cover here . ) longa and sica propose an elaborate iterated cornacchia algorithm to derive their lattice basis . in this section , we show that no lattice algorithms are necessary : we can generate four short lattice vectors using lemma [ lemma : relations ] .
recall the longa
sica construction : let @xmath312 be a glv curve , with an efficiently computable endomorphism @xmath313 , and frobenius @xmath242 . applying the gls construction ( exactly as in [ sec : gls ] ) to @xmath238 , we obtain a twisted elliptic curve @xmath251 over @xmath241 with an efficiently computable endomorphism @xmath262 corresponding to the twist of @xmath242 : if @xmath314 is the twisting isomorphism , then @xmath315 .
the endomorphisms @xmath262 and @xmath253 satisfy @xmath316 respectively . clearly , @xmath265 $ ] contains @xmath266 $ ] : the discriminants of the orders @xmath266 $ ] and @xmath265 $ ] are @xmath317 so the relative conductor of @xmath266 $ ] in @xmath265 $ ] is @xmath318 . as in vanilla gls
, we can write @xmath319 the glv endomorphism @xmath313 on @xmath238 induces a second efficient endomorphism @xmath320 on @xmath251 .
we have @xmath321 , so @xmath169 \cong { \mathbb{z}}[\phi_0]$ ] .
since @xmath313 is a glv endomorphism , @xmath169 $ ] is either the maximal order of the endomorphism algebra of @xmath251 , or very close to it
so it makes sense to assume that @xmath169 $ ] contains @xmath265 $ ] ( cf .
remark [ rem : ls - hyp ] below ) , so that we can write @xmath262 as @xmath322 where @xmath323 observe that @xmath68 and @xmath69 are both in @xmath324 . as with conventional glv curves , @xmath69 and @xmath68
are both already known as byproducts of the curve construction ( via the cm method ) or its order computation : if @xmath325 , then @xmath326 [ th : glv+gls ] with @xmath92 and @xmath262 defined as above , suppose we are in the situation of [ sec : general ] with @xmath310 .
the vectors @xmath327 generate a sublattice of determinant @xmath98 in @xmath37 . if @xmath99 , then @xmath328 .
let @xmath329 and @xmath261 be the eigenvalues of @xmath92 and @xmath262 on @xmath7 , respectively . applying lemma [ lemma : relations ] to the inclusion @xmath265 \subset { \mathbb{z}}[\phi]$ ] , we obtain relations @xmath330 corresponding to the vectors @xmath331 and @xmath332 .
multiplying the relations above through by @xmath333 and using @xmath334 , we obtain new relations @xmath335 corresponding to the vectors @xmath336 and @xmath337 , respectively .
the vectors produced by theorem [ th : glv+gls ] are short : @xmath92 is a glv endomorphism , so both @xmath119 and @xmath118 are in @xmath338 .
hence , in view of eq . , @xmath339 is in @xmath324 for @xmath340 .
[ rem : ls - hyp ] the assumption that @xmath265 $ ] is contained in @xmath169 $ ] is incompatible with the hypothesis of @xcite ( which supposes that @xmath341 ) ; but even without this assumption , @xmath342 when @xmath89 is ordinary , so the hypothesis of @xcite is _ never _ satisfied for ordinary curves .
the guillevic ionica construction @xcite uses a modified cm method to search for elliptic curves @xmath343 such that @xmath89 has endomorphisms @xmath92 and @xmath262 such that @xmath92 is separable of very small degree @xmath344 , and @xmath262 is the composition of an inseparable @xmath121-isogeny and a separable isogeny of very small degree @xmath345 .
( in a sense , these curves are to longa
sica curves what reductions of @xmath0-curves are to gls curves . )
once such @xmath89 and @xmath121 have been found ( given @xmath344 and @xmath345 ) , the @xmath92 and @xmath262 are easily recovered using vlu s formul @xcite .
if this construction is used , then ( as with the standard cm method ) the expression of @xmath87 as an element of @xmath169 $ ] is known : @xmath346 [ th : g - i ] with @xmath92 and @xmath262 defined as above : suppose we are in the situation of [ sec : general ] with @xmath310 , and @xmath347\pi$ ] .
the vectors @xmath348 generate a sublattice of determinant @xmath249 in @xmath37 . if @xmath349 , then @xmath350 .
the proof is the same as for theorem [ th : glv+gls ] , but with @xmath351 .
suppose @xmath76 is an ordinary principally polarized abelian surface ( in our applications , @xmath3 is either the jacobian of a genus 2 curve , or the weil restriction of an elliptic curve ) .
the frobenius endomorphism @xmath87 of @xmath3 generates a quartic cm field @xmath177 , and the rosati involution of @xmath352 ( exchanging an endomorphism with its rosati dual ) acts as complex conjugation on @xmath177 .
hence , the quadratic real subfield of @xmath177 is @xmath353 , and @xmath354 $ ] is a real quadratic order
. we may identify @xmath355 with @xmath356 ; the eigenvalue of @xmath357 on subgroups of @xmath358 is @xmath359 ( because @xmath87 has eigenvalue @xmath360 ) .
[ th : rm-2d ] suppose @xmath92 is a non - integer real multiplication endomorphism of an ordinary abelian surface @xmath3 ( ie , @xmath364 ) such that @xmath354\subseteq{\mathbb{z}}[\phi]$ ] , and assume that we are in the situation of [ sec : general ] with @xmath96 . if @xmath365 , then the vectors @xmath366 generate a sublattice of determinant @xmath14 in @xmath37 . if @xmath367 , then @xmath100 .
the proof is almost exactly the same as that of theorem [ th : ec-2d ] . as we noted above
, @xmath368 has eigenvalue @xmath369 on @xmath7 . applying lemma [ lemma : relations ] to @xmath354 \subset { \mathbb{z}}[\phi]$ ]
, we obtain relations @xmath370 the first implies that @xmath107 is in @xmath37 , the second that @xmath108 is in @xmath37 .
the determinant of @xmath371 is @xmath372 using eq . , @xmath373 , and @xmath374 .
we apply theorem [ th : rm-2d ] to the explicit real multiplication families treated in @xcite . in each case , the assumption @xmath375 \subseteq { \mathbb{z}}[\phi]$ ] is fulfilled because @xmath169 $ ] is the maximal order of the real subfield of the endomorphism algebra .
brumer @xcite , hashimoto @xcite , mestre @xcite , and tautz , top , and verberkmoes @xcite have given explicit constructions of families of genus 2 curves whose jacobians have explicit real multiplication by @xmath376 $ ] , which is the maximal order of @xmath377 ( see wilson s thesis @xcite for a full characterization of all such curves ) . in each case , the curve is equipped with a correspondence inducing an explicit endomorphism @xmath92 on the jacobian satisfying @xmath378 ; these endomorphisms have been exploited for fast scalar multiplication in @xcite and @xcite .
let @xmath379 be the reduction mod @xmath121 of a curve taken from one of these families ; then @xmath380 inherits the explicit endomorphism @xmath92 , and @xmath169 \cong { \mathbb{z}}[(1 + \sqrt{5})/2]$ ] . since @xmath376 $ ] is the maximal order of @xmath377 , we must have @xmath354 \subseteq { \mathbb{z}}[\phi]$ ] ; so @xmath381 where @xmath382 and @xmath383 .
mestre @xcite has constructed a two - parameter family of genus 2 curves whose jacobians have explicit real multiplication by @xmath388 $ ] ( an alternative presentation of these curves for cryptographic applications is developed in @xcite ; see bending s thesis @xcite for a full characterization of curves with rm by @xmath388 $ ] ) .
the efficient endomorphism @xmath92 satisfies @xmath389 in this case , so @xmath390 .
let @xmath379 be the reduction mod @xmath121 of a curve taken from one of these families ; @xmath380 inherits the explicit endomorphism @xmath92 , and @xmath169 \cong { \mathbb{z}}[\sqrt{2}]$ ] . since @xmath388 $ ] is the maximal order of @xmath391 , we must have @xmath354 \subseteq { \mathbb{z}}[\phi]$ ] ; so @xmath381 where @xmath382 and @xmath392 .
putting ourselves in the situation of [ sec : general ] with @xmath96 , theorem [ th : rm-2d ] yields vectors @xmath393 in @xmath37 . for this family , @xmath394 and @xmath395 ; each is much smaller than @xmath369 , so as before we have @xmath387 xx l. babai , `` on lovasz lattice reduction and the nearest lattice point problem '' .
combinatorica * 6 * ( 1986 ) 113 p. r. bending , `` curves of genus 2 with @xmath396 multiplication '' .
ph . d. thesis , university of oxford ( 1998 ) j. w. bos , c. costello , h. hisil , and k. lauter , `` high - performance scalar multiplication using 8-dimensional glv / gls decomposition '' . in g.
bertoni and j .- s .
coron , cryptographic hardware and embedded systems - ches 2013 .
lecture notes in comput .
* 8086 * ( 2013 ) 331348 a. brumer , `` the rank of @xmath397 '' .
asterisque * 228 * ( 1995 ) 4168 d. cad , x. pujol , and d. stehl , fplll .
m. ciet , f. sica , and j .- j .
quisquater , `` analysis of the gallant lambert vanstone method based on efficient endomorphisms : elliptic and hyperelliptic curves '' . in k.
nyberg and h. m. heys ( eds ) , selected areas in cryptography : sac 2002 lecture notes in comput .
* 2595 * ( 2003 ) 2136 h. cohen and g. frey ( eds . )
handbook of elliptic and hyperelliptic curve cryptography .
chapman & hall / crc ( 2006 ) s. d. galbraith , mathematics of public key cryptography .
cambridge university press ( 2012 ) s. d. galbraith , x. lin , and m. scott , `` endomorphisms for faster elliptic curve cryptography on a large class of curves '' .
j. crypt .
* 24 * # 3 ( 2011 ) 446469 r. p. gallant
, r. j. lambert , and s. a. vanstone , `` faster point multiplication on elliptic curves with efficient endomorphisms '' . in j.
kilian ( ed . ) , advances in cryptology : crypto 2001 , lecture notes in comput .
* 2139 * ( 2001 ) 190200 p.
gaudry , d. r. kohel , and b. smith , `` counting points on genus 2 curves with real multiplication '' . in d.
lee and x. wang ( eds ) , advances in cryptology : asiacrypt 2011 , lecture notes in comput .
* 7073 * ( 2011 ) 504 - 519 a. guillevic and s. ionica , `` four dimensional glv via the weil restriction '' .
to appear in asiacrypt 2013 .
k. hashimoto , `` on brumer s family of rm - curves of genus two '' .
tohoku math .
j. ( 2 ) * 52 * # 4 ( 2000 ) 475488 h. hisil , k. wong , g. carter , and e. dawson , `` twisted edwards curves revisited '' . in : j. pieprzyk ( ed . ) , advances in cryptology : asiacrypt 2008 .
lecture notes in comput .
* 5350 * ( 2008 ) 326343 m. kaib , `` the gauss lattice basis reduction algorithm succeeds with any norm '' .
in l. budach ( ed . ) , fundamentals of computation theory .
lecture notes in comput .
* 529 * ( 1991 ) 275286 d. r. kohel , `` endomorphism rings of elliptic curves over finite fields '' .
ph . d. thesis , university of california at berkeley ( 1996 ) d. r. kohel and b. smith , `` efficiently computable endomorphisms for hyperelliptic curves '' . in f.
hess , s. pauli , and m. pohst ( eds ) , algorithmic number theory : ants - vii , lecture notes in comput .
* 4076 * ( 2006 ) 495509 h. w. lenstra , jr .
, `` complex multiplication structure of elliptic curves '' .
j. number theory * 56 * # 2 ( 1996 ) 227241 p. longa and f. sica , `` four - dimensional gallant lambert vanstone scalar multiplication '' . in x.
wang and k. sako ( eds ) , advances in cryptology
asiacrypt 2012 , lecture notes in comput .
* 7658 * ( 2012 ) 718739 j .- f .
mestre , `` familles de courbes hyperelliptiques multiplications relles '' . progr .
math . * 89 * ( 1991 ) 313334 j .- f .
mestre , `` une gnralisation dune construction de richelot '' .
j. algebraic geom .
* 22 * ( 2013 ) 575580 p. l. montgomery ,
`` speeding the pollard and elliptic curve methods of factorization '' . math . comp .
* 48 * # 177 ( 1987 ) 243264 g. c. pohlig and m. e. hellman , `` an improved algorithm for computing logarithms over @xmath398 and its cryptographic significance . ''
ieee trans .
theory * 24 * ( 1978 ) 106110 h .- g .
ruck , `` abelian surfaces and jacobian varieties over finite fields '' . compositio math . *
76 * # 3 ( 1990 ) 351366 r. schoof , `` counting points on elliptic curves over finite fields '' .
j. thor .
nombres bordeaux * 7 * ( 1995 ) 219254 b. smith , `` families of fast elliptic curves from @xmath0-curves '' . to appear in asiacrypt 2013 .
http://hal.inria.fr/hal-00825287 h. p.
f. swinnerton - dyer , a brief guide to algebraic number theory .
lms student texts * 50 * , cambridge university press ( 2001 ) k. takashima , `` a new type of fast endomorphisms on jacobians of hyperelliptic curves and their cryptographic application '' .
ieice trans .
fundamentals * e89-a * # 1 ( 2006 ) 124133 w. tautz , j. top , and a. verberkmoes , `` explicit hyperelliptic curves with real multiplication and permutation polynomials '' .
j. math . * 43 * # 5 ( 1991 ) 10551064 j. wilson , `` curves of genus 2 with real multiplication by a square root of 5 '' .
ph . d. thesis , university of oxford , 1998 j. vlu , `` isognies entre courbes elliptiques '' .
c. r. math .
paris * 273 * ( 1971 ) 238241 z. zhou , z. hu , m. xu , and w. song , `` efficient 3-dimensional glv method for faster point multiplication on some gls elliptic curves '' .
process .
* 110 * # 22 ( 2010 ) 10031006 | the first step in elliptic curve scalar multiplication algorithms based on scalar decompositions using efficient endomorphisms including gallant lambert vanstone ( glv ) and galbraith lin scott ( gls ) multiplication , as well as higher - dimensional and higher - genus constructions is to produce a short basis of a certain integer lattice involving the eigenvalues of the endomorphisms .
the shorter the basis vectors , the shorter the decomposed scalar coefficients , and the faster the resulting scalar multiplication . typically , knowledge of the eigenvalues allows us to write down a long basis , which we then reduce using the euclidean algorithm , gauss reduction , lll , or even a more specialized algorithm . in this work ,
we use elementary facts about quadratic rings to immediately write down a short basis of the lattice for the glv , gls , glv+gls , and @xmath0-curve constructions on elliptic curves , and for genus 2 real multiplication constructions .
we do not pretend that this represents a significant optimization in scalar multiplication , since the lattice reduction step is always an offline precomputation but it does give a better insight into the structure of scalar decompositions . in any case , it is always more convenient to use a ready - made short basis than it is to compute a new one . |
readers of wikipedia often notice that multiple articles contain highly - similar or even identical passages . in some cases
these represent duplicate articles marked for merging , but content overlap arises in other cases as well . for example , the article about a hurricane and the article about the location where it made landfall might share the same content about the impact of the natural disaster .
identical content is most likely the result of copy and paste between articles , but interestingly , readers occasionally come across highly - similar content that state contradictory facts . in a distributed environment where anyone can edit content , these observations are perhaps not surprising , and in this paper we attempt to rigorously characterize these phenomena by treating the problem as that of near - duplicate sentence detection .
we adapt standard locality - sensitive hashing ( lsh ) techniques to identify clusters of near - duplicate sentences in wikipedia .
we believe this problem is interesting in a few ways : for duplicate sentences , our analyses quantify the extent to which wikipedia content is simply _ replicated _ , as opposed to _ written _ from scratch . in the case of _
near _ duplicates , some differences represent minor copyediting that does not change the substance of the content , but in other cases the differences represent contradictory facts . quantifying these cases
provides an indirect measure of the quality of wikipedia in terms of self consistency .
this work makes no claims about the novelty of our techniques nor our implementation in mapreduce .
rather , our contribution lies in the analysis and categorization of near - duplicate sentence types .
we have not seen locality - sensitive hashing applied to wikipedia in this way before .
the problem we tackle in this paper is related to a few other problems that have been studied before .
near - duplicate detection of web pages @xcite is important in search because web pages are often copied or mirrored with only minor differences ( e.g. , ads or navigation bars ) ; it would be desirable to return only the `` canonical '' versions in search results .
in fact , the algorithm that we use in this paper , minhash @xcite , was originally developed for exactly this purpose .
another closely - related problem is plagiarism detection @xcite , or more generally , `` text reuse '' @xcite .
in contrast to near - duplicate detection , the focus is usually on smaller segments of text as opposed to entire documents .
however , similar approaches such as shingling are applicable to both problems .
other similar formulations of the problem are what the data mining community calls pairwise similarity search or `` all pairs '' search @xcite and what the database community calls set similarity join @xcite .
the task is the same : given a potentially large collection of objects , identify all pairs whose similarity is above a threshold according to some similarity metric .
there are two classes of solutions to the above problems : in the _ index - based _ approach , an inverted index is constructed from objects in the collection and a traversal of the index allows the similar pairs to be extracted , e.g. , @xcite ; with the _ hash - based _ approach , the basic idea is to use locality - sensitive hashing ( lsh ) to identify similar pairs based on hash collisions , e.g. , minhash @xcite . of course
, hybrid solutions are also possible . scaling up these solutions
has been accomplished by mapreduce @xcite .
similarly , our approach takes advantage of minhash using a mapreduce implementation in hadoop .
because of its open nature , wikipedia has generated much controversy over its editorial quality and factual correctness .
an early study found wikipedia s accuracy to rival that of traditional encyclopedias @xcite , but subsequent investigations have arrived at conflicting conclusions .
a thorough review is beyond the scope of this short paper , but somewhat ironically , the best summary of this ongoing debate is a wikipedia article . since wikipedia may be edited anonymously , information may be freely copied between web sources and even between wikipedia articles without verification however , this is not to say that there are no quality assurance mechanisms in wikipedia @xcite .
although there are active communities of editors who contribute to the upkeep of various articles , much of wikipedia is edited and expanded in an ad hoc manner . in particular ,
wilkinson and huberman found the distribution of article edits on wikipedia to have a long tail , meaning that a small number of articles account for most of the edits , and that the number of edits is related to article quality .
articles with few edits and low editorial attention are less likely to be updated , which is a source of contradictory information based on our analysis .
for near - duplicate detection we use a well - known technique called minhash @xcite .
we begin with a parameterized family of @xmath0 hash functions @xmath1 , @xmath2 .
each sentence in a wikipedia article is broken up into @xmath3-gram `` shingles '' ( at the character level ) and for the shingle set @xmath4 a set @xmath5 of minimum hashes over the hash family is generated .
the signature of a document @xmath6 is represented as a vector of @xmath7 minhashes randomly selected from the set of @xmath0 . to increase recall
we generate @xmath8 signatures for each sentence ( i.e. , @xmath8 draws of @xmath7 from @xmath0 ) .
broder proves a straightforward relationship between minhash collisions ( i.e. , documents that share the same signature ) and their jaccard similarities , which forms the theoretical foundation of why and how minhash `` works '' ; we refer the reader to the original paper for the relevant proofs .
we implemented minhash in mapreduce @xcite using hadoop for this study .
the algorithm is as follows : each mapper receives a wikipedia article identified by a unique docid . inside the mapper we break the article into sentences using a regular expression ; sentences that are shorter than 75 shingles or longer than 600 shingles are discarded . for each sentence ,
the @xmath8 minhash signatures are then computed ( per above ) .
the family of hash functions is implemented using a `` multiply shift '' hashing scheme and generated from a random seed . for our experiments we use a 60 bit hash and a hash family of size @xmath9 .
each signature is emitted as the key of an intermediate key
value pair with the sentence i d as the value ( constructed from the docid and the sentence number ) .
the mapreduce programming model guarantees that all values associated with the same key ( minhash signatures in our case ) are shuffled to the same reducer and grouped together for further processing in effect , collecting the hash collisions for us . in the reducers
we receive signatures as keys and all sentence ids that share a signature as values .
if there is more than one value per key , we write out all sentence ids as a cluster .
this serves as input to the final cluster generation stage ( more below ) .
one complexity of applying minhash to real - world datasets is the myriad of parameters that must be selected each setting manifests a tradeoff between precision , recall , and computational effort .
our approach to parameter tuning relied on a combination of analytical calculations and hand - tuning based on examining the output .
we began by fixing the hashing scheme and the size of the hash family ( @xmath9 ) .
we then generated 10 signatures ( @xmath8 = 10 ) per input sentence .
based on broder , the probability of a match for sentences @xmath10 and @xmath11 can be expressed as follows : @xmath12 = 1 - ( 1 - s^k)^m\ ] ] where @xmath13 . with the above settings , the effects of different @xmath7 s
are shown in figure [ clust ] .
based on this analysis , we set @xmath14 .
this means that if we choose 0.9 jaccard similarity as our goal ( 90% overlap in shingle sets ) , then there is a 99% chance of a match ( i.e. , the recall ) .
finally , after some hand tuning , we settled on a shingle length of 12 characters .
this setting means that we obtain shingles that cross word boundaries , which allows us to capture ( to some extent ) word order in our similarity computations .
based on the parameter settings , it is possible to estimate the amount of data that is generated by our approach , which is the product of the length of each signature , the number of signatures per sentence , and the total number of sentences .
this computation is useful because the amount of intermediate data provides a rough proxy for algorithm running time .
finally , note that it is possible to improve precision by filtering results with a second pass on the output signature groups . in this second pass
we can discard false positives or apply another similarity metric ( e.g. , edit distance ) .
compared to the computational cost of minhash , such additional processing is cheap since it is applied to far less data .
however , we did not implement second - pass filtering in our experiments and leave this for future work .
the output of minhash is a set of clusters , where each cluster represents a signature collision .
since we generate multiple signatures per sentence , it is possible that a sentence appears in multiple clusters .
we adopt the standard practice of merging all clusters that share at least one common sentence .
cluster merging is accomplished in one pass ( outside mapreduce ) with a union find data structure @xcite . in union find each node maintains a pointer to a head node for the set .
we maintain a lookup table from each sentence i d to its node in the union find structure .
iterating over the input clusters , we look up the node for each sentence , merging clusters or creating new ones as needed . merging two clusters
involves changing the head node of one cluster to point to the head of the other .
once all clusters are processed we perform a pass over the lookup table to obtain the mapping from clusters to nodes .
another mapreduce job reconstructs the sentences in each cluster .
our analysis was conducted on an xml dump of english wikipedia from july 2013 .
the entire corpus is approximately 42 gb and contains 10.2 m articles ( after excluding certain non - article pages ) .
based on our simple sentence chunker , there are a total 135.8 m sentences . running
our hadoop implementation of minhash on the entire collection took approximately 25 minutes on our cluster consisting of 16 nodes ( 128 cores ) .
code necessary for replicating these experiments have been open sourced .
after the processing pipeline , we identified 1.15 m clusters , 3.50 m article / sentence pairs over 1.09 m articles .
the clusters contain 2.36 m unique sentences .
cluster sizes range from 2 to over 40k ; see figure [ clust ] for a histogram .
most clusters are small ; about 99% of the clusters have ten or fewer sentences , but about a quarter of the output sentence / article pairs fall into clusters with sizes greater than ten .
+ of the agricultural land 40.4% is used for growing crops and 26.6% is pastures while 2.2% is used for orchards or vine crops .
[ gondiswil ] + of the agricultural land 26.1% is used for growing crops and 30.2% is pastures while 3.0% is used for orchards or vine crops .
[ kleindietwil ] + of the agricultural land 37.8% is used for growing crops and 35.5% is pastures while 2.2% is used for orchards or vine crops .
[ leimiswil ] + * identical * + professional organizers help redirect paradigms into more useful cross - applications that ensure properly co - sustainable futures for their clients spaces and processes .
[ professional organizing ] + professional organizers help redirect paradigms into more useful cross - applications that ensure properly co - sustainable futures for their clients spaces and processes .
[ professional organizer ] + * copyediting * + in dry areas it may only emerge from its burrow for a few weeks when conditions are right and usually at night but in areas with permanent water bodies and abundant rain it may be active all day . [
great plains toad ] + in dry areas it may only emerge from its burrow for a few weeks when conditions are right and only at night but in areas with permanent water bodies and abundant rain it may be active all day .
[ list of amphibians and reptiles of montana ] + * factual drift * + bulgaria , a poor rural nation of 7 million people sought to acquire macedonia but when it tried it was defeated in 1913 in the second balkan war .
[ history of the balkans ] + bulgaria a poor rural nation of 4.5 million people sought to acquire macedonia but when it tried it was defeated in 1913 in the second balkan war . [ home front during world war i ] + * references * + komjth pter and vilmos totik : problems and theorems in classical set theory , springer - verlag , berlin , 2006 .
[ pter komjth ] + pter komjth , vilmos totik : problems and theorems in classical set theory , springer - verlag , berlin , 2006 .
[ vilmos totik ] + * other * + army medical research institute of infectious diseases ( usamriid ) microbiologist bruce e. [ timeline of the 2001 anthrax attacks ] + army medical research institute of infectious diseases ( usamriid ) which transitioned from the previous u.s .
[ fort detrick ] based on manual inspection of the cluster output , we can categorize near - duplicate clusters into one of six types .
examples are shown in figure [ toparticlegroups ] and discussed below .
_ templates _ describe sentences that have identical structure , but with different entities , facts , or figures for _ different _ topics ( and thus are not contradictory ) . they reflect conscious attempts to impose structure across groups of articles that may be related .
since several of the largest template clusters contain tens of thousands of sentences ( the largest over 40,000 ) , it is likely that some template groups are automatically generated using bots . in many cases template sentences
are found in stub articles .
_ identical _ sentences are the result of copy and paste , and are often found in articles that cover similar topics or articles that are subtopics of other topics .
non - identical but highly - similar sentences break down into two types : _ copyediting _ refers to nearly identical sentences that differ in stylistic or otherwise non - substantive ways .
they arise with minor editing after a copy and paste .
_ factual drift _ describes sentences about the _ same _ topic that provide contradictory facts .
although without detailed research , there is no way to ascertain which version ( if any ) is correct , we can identify a common scenario .
after a copy and paste , a fact becomes out of date ( e.g. , the tallest building or the death toll in a disaster ) and is corrected in one instance but not the others .
_ references _ refer to citations , typically occurring at the end of articles . since wikipedia does not adhere to one single citation style , the same work may be cited differently , or multiple citations to the same venue may be similar . finally , clusters that do not fit into any of the above categories are classified as _
other_. these typically represent sentences that are highly similar , but otherwise bear no semantic relationship with each other .
sentence chunking errors often contribute to these spurious results .
to quantify the distribution of these six cases , we randomly sampled 2094 clusters and performed manual classification .
the results are shown in table [ counts ] .
nearly three quarters of all clusters are either _ identical _ or _
templates_. from this table , we obtain a rough characterization of the precision of our analysis based on minhash except for `` other '' , all categories are `` correct '' identification of near - duplicate sentences .
recall can be computed analytically , as we have shown in the previous section .
.manual classification of sample clusters .
[ cols="<,>,>",options="header " , ] [ counts ] finally , we manually examined a few of the large clusters and noticed that they often contain a mix of different phenomena .
one common pattern is distinct groups of identical sentences , where each of the groups are near duplicates .
since there are relatively few large clusters , these nuances do not have a significant impact on the figures in table [ counts ] .
in this work , we applied minhash to the problem of near - duplicate sentence detection on wikipedia .
our mapreduce implementation is highly scalable and processes english wikipedia in a short amount of time on a modest cluster .
we found that there is a substantial amount of duplicate content in wikipedia , and that near - duplicate sentences manifest a few phenomena , the most interesting of which is contradictory facts .
how can we act upon this analysis ?
we could imagine a robot that monitors wikipedia to flag inconsistencies and requests editors to intervene and resolve .
such a service would be valuable in improving the internal consistency and quality of wikipedia .
there are several directions in which our work can be extended .
currently , our technique only identifies _
clusters _ of near - duplicate sentences missing from our analysis is the notion of information flow : which was the source article and which was the target of copying and pasting ?
are there copy `` chains '' where content was progressively copied from one article to the next , with possible `` branches '' ( and accumulation of errors along the way ) ?
our analysis of the large near - duplicate clusters suggests that there are complex edit histories that form tree - like structures .
furthermore , are there editor - specific effects ?
for example , is copying and pasting more likely by anonymous editors ? finally ,
templates represent interesting cases of near - duplicate sentences because they , in effect , encode structured knowledge .
we could formalized such knowledge in infoboxes if they are not already : it might be possible to apply information extraction techniques to generate structured linked data from template sentences .
going in the opposite direction , we might ( via a robot ) _ create _ templates sentences to impose consistency across pages in the same category to ensure that certain facts are captured in prose . | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ our study identifies sentences in wikipedia articles that are either identical or highly similar by applying techniques for near - duplicate detection of web pages .
this is accomplished with a mapreduce implementation of minhash to identify clusters of sentences with high jaccard similarity .
we show that these clusters can be categorized into six different types , two of which are particularly interesting : identical sentences quantify the extent to which content in wikipedia is copied and pasted , and near - duplicate sentences that state contradictory facts point to quality issues in wikipedia . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
fig . 1 . associated particle distribution for @xmath3 as trigger in the range @xmath42 gev / c and for @xmath3 as the associated particle in the range @xmath43 gev / c for all centralities in ( a ) d+au collisions and ( b ) au+au collisions . fig . 2 . associated particle distribution for @xmath3 as trigger in the range @xmath44 gev / c and for @xmath3 as the associated particle in the range @xmath45 gev / c for all centralities in ( a ) d+au collisions and ( b ) au+au collisions .
3 . ratios of auau(central)/auau(peripheral ) and dau(central)/dau(peripheral ) as functions of the momentum of associated @xmath3 for two trigger windows .
the data in triangles are from @xcite for auau(0 - 5%)/pp with trigger in @xmath46 gev / c .
the shaded region shows the systematic uncertainties .
the data in squares are preliminary results from phenix @xcite for dau(0 - 20%)/pp for all charged hadrons on the near side in association with pion trigger in the interval @xmath47 gev / c . | we study the pion distributions in jets produced in heavy - ion collisions in association with trigger particles restricted to definite momentum ranges .
we determine the centrality dependence of the associated particle distributions for both d+au and au+au collisions in the recombination model .
the central - to - peripheral ratios show significant dependence on centrality for au+au collisions , but negligible dependence for d+au collisions .
the results are in agreement with the qualitative features of the data . 0.5 cm pacs numbers : 25.75.dw , 25.75.gz = 6.5 in = -0.5 in oits 765.1 + july 2005 to be submitted to the + brief report section of the + phys .
rev . c. .75 cm
* rudolph c. hwa@xmath0 and zhiguang tan@xmath1 * .5 cm @xmath0institute of theoretical science and department of physics + university of oregon , eugene , or 97403 - 5203 , usa + @xmath2institute of particle physics , hua - zhong normal university , wuhan 430079 , p. r. china .5 cm in an earlier paper @xcite dihadron correlation in jets produced in heavy - ion collisions ( hic ) was studied in the recombination model @xcite . in that paper the distribution of particles ( @xmath3 at @xmath4 ) associated with a trigger ( @xmath3 at @xmath5 ) was calculated ; it is the conditional distribution for every fixed @xmath5 , denoted by @xmath6 in @xcite this equation is expressed in eq . ( 18 ) there in a form that exposes the hard parton integration .
equation ( [ 1 ] ) was then integrated over @xmath5 for trigger momentum in the range @xmath7 gev / c , corresponding to one of the trigger windows of the star experiment @xcite at the relativistic heavy - ion collider ( rhic ) . however , that integrated result does not correspond to the quantity that is presented by star as the measured associated particle distribution ( apd ) . instead of the integral of the ratio in eq .
( [ 1 ] ) , the experimental apd thus far analyzed is the ratio of the integrals @xmath8 where @xmath9 the ratio in eq .
( [ 2 ] ) was calculated in @xcite for central au+au collisions , and was shown to be in reasonable agreement with the data when due consideration of the differences in the particle types measured is taken into account . in this short paper
we show the centrality dependence of the apd in eq .
( [ 2 ] ) for both d+au and au+au collisions , as well as for two different ranges of the trigger momentum .
it should be noted that if the ranges of integration over @xmath5 in eq .
( [ 2 ] ) are taken to be infinitesimal , there is no essential difference between eqs .
( [ 1 ] ) and ( [ 2 ] ) .
there is nothing wrong with eq .
( [ 1 ] ) , which corresponds to the usual conditional probability often considered in statistical physics .
the apd in eq .
( [ 2 ] ) with a significant range of 2 gev / c for the trigger @xmath10 is preferred for data analysis mainly to gain better experimental statistics . over the intermediate @xmath10 region
where the calculated apds decrease by three orders of magnitudes , they do not differ by very much in the general shape whether eqs .
( [ 1 ] ) or ( [ 2 ] ) is used . however , the ratio of the distributions for central to peripheral collisions depends more sensitively on the trigger range .
the calculation of that ratio according to eq .
( [ 2 ] ) is the main objective of this work . as in @xcite
, we consider here only the correlation between two pions , @xmath11 , one being the trigger particle in two trigger windows , the other being the associated particle in the range @xmath12 .
this is a simpler and trackable system than what has been measured , which includes all charged hadrons , but is rich enough in the various recombination components to provide a meaningful representation of the apd . in our calculation
we use the basic formulas in the recombination model @xcite , for which @xmath13 is @xmath14 , \label{4}\end{aligned}\ ] ] where the thermal - thermal recombination component ( @xmath15 ) has been omitted , since we study here the particles in a jet . in eq .
( [ 4 ] ) @xmath16 is the distribution of hard partons in hic and is proportional to the number of binary collisions @xmath17 ; @xmath18 is the average fraction of hard partons that emerge from the dense medium to hadronize outside .
@xmath19 is the thermal parton distribution @xcite , whose parametrization for au+au collisions has now been extended to all centralities @xcite .
@xmath20 is the shower parton distribution ( spd ) given in @xcite .
@xmath21 is the fragmentation function from hard parton @xmath22 to pion and accounts for the shower - shower recombination component . for @xmath23 in eq .
( [ 2 ] ) we have @xmath24 where @xmath25 is the recombination function for the formation of a pion , and @xmath26 is 4-parton distribution , which in our case here is @xmath27 where the @xmath15 terms have been omitted for the same reason as in eq .
( [ 4 ] ) .
the details of how @xmath28 and @xmath29 are expressed in terms of spds are described in @xcite . for d+au collisions the hard parton distribution @xmath16 and
the parameters of the thermal distributions at various centralities are given in @xcite , while for au+au collisions the corresponding quantities are given in @xcite .
we now can calculate the apd in eq .
( [ 2 ] ) for d+au collisions , as well as for au+au collisions , for all centralities .
the results for @xmath30 are shown in figs . 1 and 2 for trigger windows @xmath7 gev / c and @xmath31 gev / c , respectively .
what is notable in these figures is that there is negligible centrality dependence in ( a ) for d+au collisions , but significant dependence in ( b ) for au+au collisions .
the ratios dau(cent)/dau(peri ) and auau(cent)/auau(peri ) are shown in fig . 3 as functions of the momentum of the associated particle .
the solid lines in fig .
3 are for au+au collisions and are far from 1 , indicating appreciable dependence on centrality . at @xmath32 gev
/ c the ratio is about 3 , which implies that the thermal partons in central collisions drastically enhance the production of hadrons .
the data shown in triangles are from star @xcite for auau(0 - 5%)/pp for trigger in @xmath33 gev / c with all charged hadrons included in the associated particles .
our result agree with the data for @xmath34 gev / c .
other analysis of the star data @xcite are also in support of our result .
the large factor of 3 around @xmath35 gev / c would be hard to attain by the medium modification of the fragmentation function , if fragmentation is considered instead of recombination for hadronization .
the dashed lines for d+au collisions are just above 1 , and are virtually indistinguishable from one another in the two trigger window cases .
the data shown are preliminary results from phenix on the ratio of the apds for 0 - 20% d+au to p+p collisions for all charged hadrons on the near side in association with a pion trigger in the @xmath10 range of 5 - 10 gev / c @xcite . despite the differences in the calculated and measured quantities
, the general agreement between the two suggests that our model calculation has captured the essence of the nature of the correlation .
there are also other preliminary experimental data on conditional yield in d+au collisions that show approximate independence on centrality @xcite .
the reason why there is less centrality dependence in d+au collisions is that the thermal - shower component plays a much less prominant role than it does in au+au collisions . since the parts in the factors @xmath36 in @xmath13 and @xmath23 in eq .
( [ 2 ] ) that depend on @xmath17 cancel each other , only the @xmath37 terms in eqs .
( [ 4 ] ) and ( [ 6 ] ) produce the centrality dependence , and they are minor in d+au collisions . for au+au collisions
@xmath38 recombination is insignificant in the intermdediate @xmath10 region , so the @xmath39 terms are dominant in eqs .
( [ 4 ] ) and ( [ 6 ] ) , resulting in the dependence on centrality shown in part ( b ) of figs . 1 and 2 .
figure 3 exhibits the difference of the two colliding systems in terms of ratios .
we have calculated the associated particle distributions only for @xmath40 gev / c because our formalism for particle production is not reliable for @xmath41 gev / c .
for that reason we do not integrate the associated particle distribution to determine the yield per trigger , since it is dominated by the contribution from the low @xmath10 region . nevertheless , by inspection of the distributions in figs . 1 and 2
, it is clear that the yield / trigger is not likely to depend on centrality very much for d+au collisions , but significantly for au+au collisions .
our conclusion is that when the associated particle distributions are calculated as the ratio of the integrals given in eq .
( [ 2 ] ) , we find significant dependence on centrality for au+au collisions @xcite , but negligible dependence on centrality for d+au collisions , and is in agreement with the data @xcite .
this work was supported , in part , by the u. s. department of energy under grant no .
de - fg03 - 96er40972
. 99 r. c. hwa and c. b. yang , phys .
rev .
c * 70 * , 054902 ( 2004 ) .
r. c. hwa and c. b. yang , phys .
rev .
c * 70 * , 024905 ( 04 ) .
star collaboration , c. adler _ et al .
_ , phys .
rev .
lett . *
90 * , 082302 ( 2003 ) ; j. adams _ et al .
, ibid .
_ * 91 * , 072304 ( 2003 ) ; nucl - ex/0501016 ; f. wang , j. phys .
g * 30 * , s1299 ( 2004 ) .
r. c. hwa and z. tan , nucl - th/0503052v2 , phys .
rev .
c ( to be published ) .
r. c. hwa and c. b. yang , phys .
rev .
c * 70 * , 024904 ( 2004 ) .
r. c. hwa and c. b. yang , phys .
rev .
lett . * 93 * , 082302 ( 2004 ) .
j. adams et al ( star collaboration ) , nucl - ex/0501016 .
y. guo ( for star collaboration ) , riken bnl research center workshop on jet correlations at rhic " , march 2005 .
n. grau ( for phenix collaboration ) , riken bnl research center workshop on jet correlations at rhic " , march 2005 .
j. bielcikova ( for star collaboration ) , riken bnl research center workshop on jet correlations at rhic " , march 2005 . |
the standard quickest change detection problem is set to detect some unknown time point at which certain signal probability distribution changes over a sequence of observations .
recently , with the development of wireless sensor networks , multiple sensors can be deployed to execute the quickest change detection , and the sensors can send quantized or unquantized observations or certain local decisions to a control center , who then makes a final decision @xcite . most of the existing work is based on an assumption that the statistical properties of observations at all sensors change simultaneously . however , in certain scenarios , this assumption may not hold well .
for instance , when multiple sensors are used to detect the occurrence of the chemical leakage , the sensors that are closer to the leakage source usually observe the change earlier than those far away from the source . in such cases , two interesting problems arise : one is to detect the change as soon as possible ; the other is to identify which sensor is the closest to the source , such that we could have a first - order inference over the leakage source location .
as far as we know , currently there are few work studying the case of change occurring non - simultaneously . in the related work , the authors in @xcite proposed a scheme that each sensor makes a local decision with the computing burden at the local sensors , where they did not consider the identification problem . in @xcite , the authors modeled the change propagation process as a markov process to derive the optimal stopping rule and assumed that the change pattern always first reaches a predetermined sensor , such that the identification problem is ignored . in @xcite ,
the identification problem for the special case of two sensors was studied , where the sufficient statistic is proven as a markov process and a joint optimal stopping rule and terminal decision rule are proposed . in this paper
, we study the joint change point detection and identification problem over a linear array of @xmath0 sensors , where the change first occurs near an unknown sensor , then propagates to sensors further away .
we assume that all sensors send their observations to a control center . with the sequential observation signals , the control center first operates a stopping rule to decide when to alarm that the change has occurred
; then the control center deploys a terminal decision rule to determine which sensor that the change pattern reaches first . in our setup , three performance metrics are of interest : i ) detection delay , which is the time interval between the moment that the change occurs and the moment that an alarm is raised ; ii ) false alarm probability , which is the probability that an alarm is raised before the actual change occurs ; and iii ) false identification probability , which is the probability that the control center does not correctly identify the sensor that the change pattern first reaches .
we apply the markov optimal stopping time theory to design the optimal decision rules to minimize a weighted sum of the above three metrics .
furthermore , we derive a scheme with a much simpler structure and certain performance guarantee . the rest of this paper is organized as follows . in section [ sec :
model ] , we introduce the system model . in section [ sec : solution ] , we derive the optimal decision rules . in section [ sec : approx ] , we propose a scheme approximate to the optimal decision rules with a much lower complexity . in section
[ sec : num ] , we present some numerical results , with conclusions in section [ sec : con ] .
we consider a scenario with @xmath0 sensors constructing a linear array to monitor the environment , as shown in fig .
[ array ] . at an unknown time point ,
a change occurs at an unknown location and propagates , where we use change point time @xmath1 to denote the time that the change pattern reaches sensor @xmath2 .
we further use @xmath3 to denote the index of the sensor that the change pattern first reaches .
we focus on the bayesian setup and use @xmath4 to denote the prior probability of @xmath5 , simply with @xmath6 . conditioned on the event that the change pattern first reaches sensor @xmath2 ,
@xmath1 is assumed to bear a geometric distribution @xcite with parameter @xmath7 , @xmath8 i.e. , @xmath9=\rho(1-\rho)^k , k\geq0,\ ] ] where @xmath10 denotes the discretized time and takes integer values .
we consider the practical factors in the environment , such as the wind or the blockers , which will affect the propagation speed of the change .
for instance , see in fig .
[ array ] , if the direction of the wind is from the left to the right side in the monitored scenario , then the propagation of the air pollution will be much faster at the right side of sensor @xmath11 than that of the left side . and at the same side
, the propagation follows the deterministic order shown as @xmath12 we further assume that after the change patten reaches the first sensor @xmath2 , for the right side of sensor @xmath2 , it will propagate from one sensor to another sensor following the geometric propagation models as @xmath13=\rho_1(1-\rho_1)^{k_{2}},j > i , k_2\geq0,\ ] ] while for the left side of sensor @xmath2 , the propagation follows @xmath14=\rho_2(1-\rho_2)^{k_{2}},j
< i , k_2\geq0,\ ] ] where @xmath15 and @xmath16 are used to model possibly different propagation speed along each direction , e.g. , @xmath17 means the propagation speed is higher at the right side that that of the left side .
taking above assumption , for @xmath11 , we define all possible @xmath18 events at time @xmath10 as follows : @xmath19 where @xmath20 denote the events that after the change pattern first reaching sensor @xmath2 , it propagates across the sensors sequentially at the right side of sensor @xmath2 .
the number of events equals to the number of sensors at the right side plus 1 , i.e. , @xmath21 . @xmath22
denote the events that after the change pattern reaching sensor @xmath2 and @xmath23 , it propagates across the sensors sequentially at the right side of sensor @xmath2 , and the number of events is also @xmath21 , which is the same for the case that after the change pattern reaching sensor @xmath2 , @xmath23 , @xmath24 and so on .
since there are @xmath23 sensors at the left side of sensor @xmath2 and the event of no change pattern reaches any sensor is @xmath25 , the total number of possible events is @xmath18 . at each time @xmath10
, we assume that the observations @xmath26 $ ] from all sensors are available at a control center . for each sensor @xmath2 , conditioned on @xmath1 ,
@xmath27 is identically and independently distributed ( iid ) according to @xmath28 before @xmath1 , and iid according to @xmath29 after @xmath1 , i.e. , @xmath30 the observation sequence @xmath31 generates a filtration @xmath32 with @xmath33 where @xmath34 denotes the smallest @xmath35-field in which @xmath36 is measurable and @xmath37 .
we use @xmath38 to denote the probability measure that specifies the prior distribution of @xmath5 , the distribution of change point time , and the distribution of @xmath31 .
we also use @xmath39 to denote the expectation under the probability measure @xmath38 .
specifically , we use @xmath40 to denote the probability measure when @xmath11 . with above setups , the control center needs to detect the earliest change point time @xmath37 as soon as it occurs .
a stopping time @xmath41 will be decided for when to stop sampling and alarm that a change has occurred , where a false alarm may happen if @xmath42 .
we target to minimize the averaged detection delay @xmath43 with keeping the false alarm probability @xmath44 $ ] small .
in addition , we also require the control center to identify which sensor the change pattern reaches first .
we adopt @xmath45 to denote the @xmath46-measurable terminal decision rule used by the control center to make the identification , and @xmath47 to denote the index of the sensor identified , i.e. , @xmath48 .
a false identification occurs if @xmath49 , such that we also want to keep @xmath50 $ ] small .
we use @xmath51 to denote the sequence of terminal decision rules . summarizing above ,
our goal is to design a stopping time @xmath41 and a terminal decision rule @xmath52 that minimize the aggregated risk function defined as @xmath53+c_1\mathbb{e}\{(\tau-\gamma)^{+}\}+c_2p[\hat{b}_1\neq b_1],\ ] ] where @xmath54 and @xmath55 are appropriate constants that balance the three costs .
, scaledwidth=50.0% ]
with the posterior probabilities defined above , we denote @xmath63 .
we first have the following theorem regarding the optimal terminal decision rule @xmath52 . for any stopping time @xmath41 ,
the optimal terminal decision rule is @xmath64 and we have @xmath65=\mathbb{e}\left\{1-\max\left\{p_\tau^{1}, ... ,p_\tau^{n}\right\}\right\}.\end{aligned}\ ] ] the proof follows from proposition 4.1 of @xcite .
theorem 1 implies that the optimal terminal decision rule is simply to choose the sensor that has the largest posterior probability .
a similar situation also arises in the multiple hypothesis testing problem considered in @xcite . using above optimal terminal decision rule
, we can further express the optimization objective in as a function of the posterior probabilities defined in ( [ def_pi ] ) and ( [ def_p ] ) , as shown below .
[ lem : cost ] for any stopping time @xmath41 , can be written as @xmath66 based on the bayesian s rule , we have @xmath67 = p[t_{0,k}\left| { { { \cal f}_k } } \right . ] & = \sum\limits_{i = 1}^n { p[t_{0,k}\left| { { { \cal f}_k } } \right.,b = i ] } p_k^i \nonumber\\ & = \sum\limits_{i = 1}^n { { \pi _ { 0,k\left| i \right . } } } p_k^i.\end{aligned}\ ] ] further , according to proposition 5.1 in @xcite , @xmath68+c_1\mathbb{e}\{(\tau-\gamma)^{+}\}\nonumber\\ & = \mathbb{e}\left\{p[\tau<\gamma|\mathcal{f}_\tau]+c_1\sum \limits_{k=0}\limits^{\tau-1}p[\gamma\leq k|\mathcal{f}_k]\right\}\nonumber\\ & = \mathbb{e } \left\ { \sum\limits_{i = 1}^n { { \pi _ { 0,\tau \left| i \right . } } } p_\tau ^i+ { c_1}\sum\limits_{k = 0}^{\tau - 1 } \left(1 - \sum\limits_{i = 1}^n { { \pi _ { 0,k\left| i \right . } } } p_k^i \right)\right\}.\end{aligned}\ ] ] by combining ( [ decisionrule1 ] ) , we complete the proof .
furthermore , we have the following lemma regarding @xmath69 .
there is a time - invariant function @xmath70 such that @xmath71 .
we have @xmath72\nonumber\\ & = \frac{{f({{\bf{z}}_k}\left| { { { \cal f}_{k - 1}},t_{j , k } } \right.,{s } = i)p[t_{j , k}\left| { { { \cal f}_{k - 1}},s = i } \right.]}}{{\sum\limits_{j = 0}^{i(n -i+ 1 ) } { f({{\bf{z}}_k}\left| { { { \cal f}_{k - 1}},t_{j , k } } \right.,{s } = i)p[t_{j , k}\left| { { { \cal f}_{k - 1}},s = i } \right . ] } } } , \end{aligned}\ ] ] in which @xmath73 for another element @xmath78 in @xmath69 , we have @xmath79 \\ & = \frac{{f({{\bf{z}}_k}\left| { { { \cal f}_{k - 1}},{s } = i } \right.)p[{s } = i\left| { { { \cal f}_{k - 1 } } } \right.]}}{{\sum\limits_{n = 0}^{m-1 } { f({{\bf{z}}_k}\left| { { { \cal f}_{k - 1}},{s } = n } \right.)p[{s } = n\left| { { { \cal f}_{k - 1 } } } \right . ] } } } , \end{split}\ ] ] where @xmath80,\end{aligned}\ ] ] which can be calculated by using ( [ direcltywrite ] ) and ( [ transition ] ) .
hence , @xmath78 can also be computed by @xmath76 and @xmath77 .
this lemma implies that the posterior probabilities @xmath69 can be recursively computed from @xmath76 and @xmath77 .
combined with lemma [ lem : cost ] , we know that @xmath69 is a sufficient statistic for the problem of minimizing ( [ formulation ] ) . thus , the problem at the hand is a markov stopping time problem .
therefore , we could borrow results from the optimal stopping time theory to design the optimal decision rules for our problem .
we first consider a finite time horizon case , in which one has to make a decision before a deadline @xmath81 , i.e. , @xmath82 .
it is easy to check that the cost - to - go functions are @xmath83 where @xmath84 applying the optimal stopping time theory @xcite , we have the following theorem for the optimal decision rules .
[ thm2 ] the optimal stopping time is obtained as @xmath85 with the optimal terminal decision rule is given in ( [ decisonrule ] ) .
in the infinite time horizon case when @xmath86 , we have @xmath87 defined as @xmath88 since we have @xmath89 , @xmath90 , and the fact that all strategies allowed with deadline @xmath81 are also allowed with deadline @xmath91 .
since the observations are memoryless and conditionally iid , @xmath87 is the same for all @xmath10 ; we then use @xmath92 to denote @xmath87 .
thus , @xmath93 is derived as @xmath94 in which the interchange of @xmath95 and @xmath96 is allowed due to the dominated convergence theorem .
therefore , when the deadline is infinite , the optimal stopping rule becomes @xmath97 with the optimal terminal decision rule is given in ( [ decisonrule ] ) .
when @xmath0 is large , the optimal stopping rule does not have a simple structure , which makes the implementation highly costly . in this section ,
we propose a much simpler rule which approximates to the optimal stopping rule .
[ lemma3 ] the sequence @xmath98 is a supermartingale , i.e. , @xmath99 the proof follows from page 477 of @xcite , by using fatou s lemma .
we can use lemma [ lemma3 ] to derive the following approximation of the optimal stopping rule . in the asymptotic case of the rare change occurring with @xmath100 , one approximation of the optimal stopping rule has the following simple structure @xmath101 where @xmath102 .
and we use the optimal terminal decision rule specified in ( [ decisonrule ] ) .
for the second part of ( [ a(t-1 ) ] ) , according to lemma [ lemma3 ] , @xmath106f(\mathbf{z}_t|\mathcal{f}_{t-1})d\mathbf{z}_t \leq c_2(1-p_{t-1}^{i_{t-1}}).\ ] ] plugging the above two results ( [ firstpart ] ) and ( [ p2 ] ) into ( [ a(t-1 ) ] ) , we have @xmath107 in the sequel , we assume that @xmath108 equals to the right side of ( [ inequ ] ) . according to ( [ cost - to - go ] ) , we have if @xmath109 , @xmath110 if @xmath111 , @xmath112 we define the following transformation as @xmath113 then @xmath114 further we have @xmath115 and @xmath116 then , @xmath117 can be rewritten as @xmath118 we define @xmath119 and @xmath120 as @xmath121 @xmath122 then straightly we see that @xmath123 , @xmath124 , and @xmath125\mathbb{i}\left(\left\{\sum\limits_{j = 1}^{m-1 } { v_{t-1,j } } \leq \frac{1}{c_1}\right\}\right).\ ] ] for the next steps , we follow the proof of theorem 2 of @xcite , which is skipped here . and
additionally we use lemma [ lemma3 ] .
finally , it can be derived that @xmath126 and the test structure reduces to stopping when @xmath127 therefore , we have the structure of the stopping rule as stated in theorem 3 .
regarding to theorem 3 , we have several notes as follows .
\1 ) from lemma [ lemma3 ] and theorem [ thm2 ] , we see that @xmath128 is a lower bound of the optimal stopping time , i.e. @xmath129 , in the case of @xmath100 .
the supermartingale property shown in lemma [ lemma3 ] plays an important role in deriving @xmath128 .
the tightness of this lower bound is related to the relationship between @xmath130 and @xmath131 .
the simulation results in section [ sec : num ] show that @xmath132 and @xmath131 are quite close , which indicates that @xmath128 would be close to @xmath133 .
\2 ) from ( [ eq1_lm1 ] ) and ( [ v_kl ] ) , we have the testing statistic @xmath134 as @xmath135}{\rho p[\gamma > k\left| { { { \cal f}_k } } \right.]}.\end{aligned}\ ] ] this structure conforms to the well - known shiryaev s procedure @xcite , which is the optimal stopping rule for single sensor with iid observations and bayesian setting .
given that it is hard to efficiently compute the solution structure in ( [ stoprule ] ) , we compute the approximate optimal stopping rule in ( [ appro_stoppingrule ] ) and simulate its performance .
we assign 5 nodes constructing a linear sensor array and assume that @xmath136 and @xmath137 .
the change point time is generated according to the geometric distribution with @xmath138 , @xmath139 and @xmath140 , respectively . according to ( [ eq1_lm1 ] ) , the false alarm probability with @xmath128 is @xmath141=\mathbb{e}\left\{{\sum\limits_{i = 1}^n{{\pi _ { n+1,\tau_{app}\left| i \right . } } } p_{\tau_{app}}^i}\right\}\leq\frac{c_1}{c_1+\rho}=\alpha.\ ] ] thus we have @xmath142 , where @xmath143 is the maximum allowance for the false alarm probability , which could determine the required select @xmath54 value . in fig .
[ pfavsadd ] , we illustrate the relationships among the false alarm probability , the false identification probability , and the averaged detection delay .
we see that as the averaged detection delay increases , the false alarm probability decreases .
when the averaged detection delay becomes large , the false identification probability does not decrease much and a probability floor appears , which is due to the fact that only the samples between the time when the change pattern reaches the first sensor and the time when it reaches the second sensor can be used to effectively distinguish the sensor that the change pattern first reaches . since this part of the samples is limited , which will not increase with the detection delay , a false identification probability floor exists . in fig .
[ p_figure ] , we draw the posterior probability @xmath144 over time , where we assume that the change pattern first reaches node 3 , and then propagates to node 4 . we see that as time goes , @xmath145 gradually becomes larger than the others , which indicates that node 3 should be identified . in fig .
[ bound_figure ] , we show the relation between @xmath146 and @xmath147 in ( [ p2 ] ) .
since ( [ p2 ] ) is the key in deriving the our simplified rule , the fact that these two curves are close suggests that the performance of our low - complexity rule might be close to that of the optimal stopping rule in ( [ stoprule ] ) and ( [ stoprule_infi ] ) . vs. @xmath148
we have studied the quickest change point detection problem and the closest - node identification problem over a sensor array .
we have proposed an optimal decision scheme combing the stopping rule and the identification rule to alarm the change happening and to determine the sensor closest to the change source .
since the structure the obtained optimal scheme is complex and impractical to implement , we have further proposed a scheme with a much simpler structure . | in this paper , we consider the problem of quickest change point detection and identification over a linear array of @xmath0 sensors , where the change pattern could first reach any of these sensors , and then propagate to the other sensors .
our goal is not only to detect the presence of such a change as quickly as possible , but also to identify which sensor that the change pattern first reaches .
we jointly design two decision rules : a stopping rule , which determines when we should stop sampling and claim a change occurred , and a terminal decision rule , which decides which sensor that the change pattern reaches first , with the objective to strike a balance among the detection delay , the false alarm probability , and the false identification probability .
we show that this problem can be converted to a markov optimal stopping time problem , from which some technical tools could be borrowed . furthermore , to avoid the high implementation complexity issue of the optimal rules
, we develop a scheme with a much simpler structure and certain performance guarantee . |
the equation of state of strongly interacting matter is of paramount interest for understanding and describing the dynamics of relativistic heavy - ion collisions and the early universe and compact stellar objects as well .
the grand canonical potential @xmath1 depends on a set of intensive quantities ( temperature @xmath2 , various chemical potentials accumulated in @xmath3 ) , and parameters like quark masses , flavor number @xmath4 and color number @xmath5 . for qcd as theory of hadrons and quarks and gluons these parameters
are fixed , but it is instructive to study their impact on @xmath1 , given the complexity of qcd .
of particular interest is @xmath1 near the demarcation line ( phase boundary ) @xmath6 of hadrons vs. quarks and gluons .
this region is probed in heavy - ion collisions @xcite and is traversed in the cosmic evolution @xcite . to parameterize the equation of state near @xmath0
we employ a quasiparticle model ( qpm ) @xcite which is adjusted to available lattice qcd data .
the striking simplicity of the model consists in the expressions for net baryon density @xmath7 and entropy density @xmath8 reading @xmath9 @xmath10 + \epsilon_i\frac{(\omega_i-\mu_i)/t}{e^{(\omega_i-\mu_i)/t}+\epsilon_i } + [ \mu_i \rightarrow -\mu_i ] \right ) \
, , \label{eq.2}\ ] ] where @xmath11 , @xmath12 , @xmath13 , @xmath14 and @xmath15 . the pressure @xmath16 , where @xmath17 is the volume , is to be calculated consistently with eqs .
( [ eq.1 ] , [ eq.2 ] ) , cf . @xcite .
we restrict here our attention onto considering one independent chemical potential @xmath18 .
@xmath19 are the dispersion relations for the quasiparticle excitations with self - energies @xmath20 in line with 1-loop approximations and a procedure to include nonzero quark masses according to @xcite @xmath21 where @xmath22 . in the quark sector , the rest masses @xmath23 contain a true rest mass term @xmath24 and an artificial lattice
mass term @xmath25 introduced for calculational purposes on the lattice , @xmath26 .
the interaction generates the self - energy contributions @xmath27 with @xmath28 , @xmath29 , @xmath30 and @xmath31 .
@xmath32 is an effective coupling strength parametrized at @xmath33 via @xmath34 \!\ !
g^2_{\rm 2-loop}(\zeta(t_c ) ) + b \left(1{-}\frac{t}{t_c}\right ) , \ t{\,<\,}t_c \end{array } \right.\ ] ] with @xmath35 , and approaches smoothly the perturbative region at large temperatures .
near @xmath0 , @xmath32 becomes large , and the shift parameter @xmath36 regulates the coupling .
below @xmath0 , the coupling changes drastically its behavior . the model may be considered as a resummed expression for the thermodynamic potential , as a formal power expansion in @xmath32 generates an infinite series of terms , including also a term proportional to the plasmon term , for instance .
it goes beyond a perturbative expansion scheme as result of the effective coupling @xmath32 , which also may repair possible shortcomings of the 1-loop inspired parameterizations of the dispersion relations . in this form
, the model does not contain critical point ( cf .
@xcite ) or color - flavor locking effects ( cf .
@xcite ) . despite of its simplicity
, the equation of state for real and imaginary chemical potential as well as various susceptibilities are described very well @xcite in the qpm . here
, we describe one new aspect of our model , namely a naive chiral extrapolation of the equation of state .
we emphasize also the capability to extrapolate lattice qcd results from imaginary @xmath18 to real @xmath18 .
we focus on the region @xmath37 , where interaction effects are strong .
it is the region of the strongly coupled quark - gluon plasma @xcite presently explored experimentally at rhic and in near future at lhc and later on at fair .
the qpm parametrization of the lattice qcd results @xcite for @xmath38 with fairly large quark masses , i.e. @xmath39 and @xmath40 , was already presented in @xcite for the scaled pressure @xmath41 at @xmath42 .
the qpm parameters read @xmath43 , @xmath44 , @xmath45 for @xmath46 .
now we try to accomplish a chiral extrapolation by means of eq .
( [ equ : self ] ) .
neglecting naively a conceivable dependence of @xmath47 , @xmath48 and @xmath49 on @xmath23 , the extrapolated results are exhibited in the left panel of figure [ fig : pnf2 + 1extrap ] ( dashed curve ) for @xmath50 and @xmath51 , corresponding to the set - up in @xcite .
these lattice qcd results @xcite ( squares ) are astonishingly well reproduced , however , by the price of changing the pressure integration constant , @xmath52 , which needs to be readjusted in order to generate the small pressure below @xmath0 .
we note that putting @xmath53 in the strikt chiral limit does not change noticeably the dashed curve on the scale exhibited in figure 1 .
another important test @xcite of the suitability of our hydrodynamic eos is the interaction measure @xmath54 , where @xmath55 denotes the energy density .
as shown in the right panel of figure [ fig : pnf2 + 1extrap ] , the qpm for almost physical quark masses ( i.e. , @xmath50 and @xmath51 ) faithfully reproduces corresponding lattice qcd data @xcite of @xmath56 .
the peak , which is related to the softest point in the qcd equation of state , is located at @xmath57 . for larger temperatures ,
the interaction measure approaches logarithmically zero according to the temperature dependence in the effective coupling @xmath32 , though , is close to the conformal limit @xmath58 already for @xmath59 .
in this section , qcd thermodynamics for @xmath60 at nonzero imaginary chemical potential @xmath61 is considered .
the net quark number density @xmath62 reads in the qpm @xcite @xmath63 which is purely imaginary and an odd function in @xmath64 . in figure
[ fig : densnqimmu ] ( left panel ) , qpm results for @xmath65 , employing the parametrization from @xcite , are compared with the lattice qcd results @xcite . in particular , the pronounced bending for @xmath66 close to @xmath67 , which signals the onset of the first - order roberge - weiss transition @xcite , is accurately reproduced representing a sensible test of the qpm at nonzero @xmath7 . in the qpm , the bending is driven by the increasing @xmath64 dependence in the quasiparticle dispersion relation @xmath68 close to @xmath69 . for temperatures
@xmath70 we observe an independence of @xmath71 considered as a function of @xmath72 on the explicit value of @xmath2 .
this independence follows from eq .
( [ equ : densquarkimmu2 ] ) as long as @xmath68 is approximately independent of @xmath64 .
+ the phase boundary @xmath73 for @xmath60 is shown in the right panel of figure [ fig : densnqimmu ] for imaginary as well as for real baryo - chemical potential @xmath74 , which is accessible by analytic continuation . within the qpm ,
an estimate for @xmath73 follows as self - consistent solution of a partial differential equation based on maxwell s relation emerging at @xmath75 ( cf .
@xmath73 for imaginary @xmath76 and the first roberge - weiss transition line cross each other at @xmath77 and @xmath78 being close to the lattice qcd results @xmath79 and @xmath80 @xcite
. the features of the phase diagram in the imaginary chemical potential sector can be described as follows .
dashed curves represent the analytic sections of the first two roberge - weiss transition lines @xmath81 ( here @xmath82 ) turning into first - order transitions ( solid section ) while the dotted curve shows the first @xmath83 center symmetry line @xmath84 .
the roberge - weiss periodicity @xcite implies that sectors between @xmath85 and @xmath86 ( sector ii for @xmath87 ) are repeated copies of sector i between @xmath88 and @xmath89 .
the subsector between @xmath90 and @xmath89 ( sector i ) is a reflected copy of the subsector between @xmath88 and @xmath91 ( sector i ) mirrored at the first roberge - weiss transition line .
in summary we show that the quasiparticle model with the chosen dispersion relations for quarks and gluons accounts fairly well for the quark mass dependence in the qcd equation of state near @xmath0 as delivered by selected lattice qcd results .
we emphasize further that lattice qcd calculations at imaginary chemical potential , avoiding the sign problem of the fermion determinant , give valuable information which can be continued to real chemical potential within our model .
the feasibility of exploring larger net baryon densities is relevant for future heavy - ion experiments at fair .
+ the work is supported by 06dr136 , gsi - fe , and eu i3hp .
999 m. bluhm , b. kmpfer , r. schulze , d. seipt , and u. heinz , phys .
c * 76 * , 034901 ( 2007 ) .
h. schade , and b. kmpfer , arxiv:0705.2003 .
a. peshier , b. kmpfer , o. p. pavlenko , and g. soff , phys .
b * 337 * , 235 ( 1994 ) .
a. peshier , b. kmpfer , o. p. pavlenko , and g. soff , phys .
d * 54 * , 2399 ( 1996 ) .
a. peshier , b. kmpfer , and g. soff , phys .
c * 61 * , 045203 ( 2000 ) .
a. peshier , b. kmpfer , and g. soff , phys .
d * 66 * , 094003 ( 2002 ) .
m. bluhm , b. kmpfer , and g. soff , phys .
b * 620 * , 131 ( 2005 ) .
m. bluhm , b. kmpfer , r. schulze , and d. seipt , eur .
j. c * 49 * , 205 ( 2007 ) .
r. d. pisarski , nucl .
a * 498 * , 423c ( 1989 ) .
f. karsch , lect .
notes phys .
* 583 * , 209 ( 2002 ) .
d. h. rischke , prog .
phys . * 52 * , 197 ( 2004 ) .
m. bluhm , and b. kmpfer , phys .
d * 77 * , 114016 ( 2008 ) .
m. bluhm , and b. kmpfer , phys .
d * 77 * , 034004 ( 2008 ) .
e. v. shuryak , prog .
. phys . * 53 * , 273 ( 2004 ) ; nucl .
a * 750 * , 64 ( 2005 ) .
m. gyulassy , and l. mclerran , nucl .
a * 750 * , 30 ( 2005 ) .
d. a. teaney , phys .
c * 68 * , 034913 ( 2003 ) ; nucl .
phys . a * 785 * , 44 ( 2007 ) .
f. karsch , k. redlich , and a. tawfik , eur .
j. c * 29 * , 549 ( 2003 ) .
m. cheng et al . , phys .
d * 77 * , 014511 ( 2008 ) . c. m. hung , and e. v. shuryak , phys .
lett . * 75 * , 4003 ( 1995 ) .
m. delia , and m .-
lombardo , phys .
d * 70 * , 074509 ( 2004 ) .
m. delia , f. di renzo , and m .-
lombardo , phys .
d * 76 * , 114509 ( 2007 ) . m. delia , and m .-
lombardo , phys .
d * 67 * , 014505 ( 2003 ) .
a. roberge , and n. weiss , nucl .
b * 275 * , 734 ( 1986 ) . | 1.0 cm 1.0 cm we test the quark mass dependence implemented in the quasiparticle dispersion relations of our quasiparticle model for the qcd equation of state by comparing with recently available lattice qcd data near @xmath0 employing almost physical quark masses .
in addition , we emphasize the capability of our model to successfully describe lattice qcd results for imaginary chemical potential and to analytically continue the latter to real chemical potential . |
precision observations of the cosmic microwave background @xcite , the large scale structure of galaxies @xcite , and distant type ia supernovae @xcite have led to a new standard model of cosmology in which the energy density is dominated by dark energy with negative pressure , leading to an accelerated expansion of the universe .
a very interesting proposal is the so - called mass varying neutrino ( mavan ) model @xcite in which a light scalar field couples to neutrinos , see also @xcite for background .
because of the coupling , the mass of the scalar field does not have to be as small as the hubble scale but can be larger , while the model still accomplishes late - time acceleration . in this paper
we discuss the different criteria that need to be fulfilled in order to have a stable , adiabatic mass varying neutrino model with the correct cosmology today - i.e. @xmath0 , @xmath1 and @xmath2 .
our aim is to show that it is only possible as long as the scalar field potential resembles a cosmological constant - and hence the model loses its prediction of the current neutrino mass and fails to solve the coincidence problem .
in addition , we discuss the suggestion that the lightest neutrino , which can be relativistic today , may be responsible for dark energy .
we find that there is evidence that the relativistic neutrino will feel an instability towards the formation of neutrino nuggets . in the next section
we briefly review the formalism needed to study mass varying neutrinos and in sec .
[ sec:3 ] we discuss the different criteria in the mavan framework . in sec .
[ sec:4 ] we discuss mavans with a relativistic neutrino and in sec . [ sec:5 ] we conclude .
the idea in the mass varying neutrino scenario @xcite is to introduce a coupling between neutrinos and a light scalar field and to identify the coupled fluid with dark energy . in this scenario
the neutrino mass @xmath3 is generated from the vacuum expectation value ( vev ) of the scalar field .
thus at scale factor @xmath4 the pressure @xmath5 and energy density @xmath6 of the uniform neutrino background contribute to the effective potential @xmath7 of the scalar field in the following way @xmath8 where @xmath9 denotes the fundamental scalar potential .
the energy density and pressure of the scalar field are given by the usual expressions , @xmath10 defining @xmath11 to be the equation of state of the coupled dark energy fluid , where @xmath12 denotes its pressure and @xmath13 its energy density , the requirement of energy conservation gives , @xmath14 here @xmath15 and we use dots to refer to the derivative with respect to conformal time . taking eq .
( [ eq : econserv ] ) into account , one arrives at a modified klein - gordon equation describing the evolution of @xmath16 , @xmath17 here and in the following primes denote derivatives with respect to @xmath16 ( @xmath18 ) and @xmath19 is the coupling between the scalar field and the neutrinos . in the nonrelativistic case @xmath20 ,
such that eq .
( [ eq : eff ] ) takes the form @xmath21 assuming the curvature scale of the potential and thus the mass of the scalar field @xmath22 to be much larger than the expansion rate of the universe , @xmath23 the adiabatic solution to the equation of motion of the scalar field in eq .
( [ eq : kg ] ) applies @xcite . as a consequence
, the scalar field instantaneously tracks the minimum of its effective potential @xmath24 @xmath25 mavan models has a possibility of becoming unstable on sub - hubble scales @xmath26 in the nonrelativistic regime of the neutrinos , where the perturbations @xmath27 evolve adiabatically . in ref .
@xcite it is shown that the equation of motion for the neutrino density contrast @xmath28 in the regime @xmath26 can be written as @xmath29\end{aligned}\ ] ] where @xmath30 since neutrinos not only interact through gravity , but also through the force mediated by the scalar field , they feel an effective newton s constant @xmath31 as defined in eq .
( [ geff ] ) .
the force depends upon the mavan model specific functions @xmath32 and @xmath33 and takes values between @xmath34 and @xmath35 on very large and small length scales , respectively . in certain cases of strong coupling neutrinos
suffer an instability towards clumping in which case they stop behaving as dark energy @xcite . in ref .
@xcite a criterion for the stability was developed .
this can be recast in a more convenient form @xmath37 , where we have neglected the effect of baryons compared to cold dark matter and we have assumed the density contrasts of roughly the same order .
in summary we have the following different criteria for an adiabatic mavan model . *
the model needs to satisfy current observations i.e. @xmath2 .
this is easily fulfilled by demanding @xmath38 since we know the neutrino contribution to both pressure and energy density is smaller than the scalar field contribution . *
we want a neutrino mass at present that satisfies the current neutrino mass bounds from cosmology , that is @xmath39 . using this
, we get a maximal value for the neutrino density parameter @xmath40 @xcite . *
the model must produce the right cosmology , i.e. @xmath41 .
this gives us roughly @xmath42 .
* in order to have a prediction for the current neutrino mass , we are looking for an adiabatic model that continuously tracks the minimum of the effective potential , i.e. @xmath43 .
in addition , to avoid severe fine - tuning , we demand @xmath44 @xcite .
hence @xmath45 , using a constant coupling gives us @xmath46 .
* we want our model to be stable towards the formation of neutrino nuggets at present .
this gives the requirement @xmath37 .
since we are in the adiabatic regime , we are looking on scales where @xmath47 . combined with the fact that @xmath48 , this criterion simply reduces to @xmath49 . combining the criteria above and using @xmath50
, we can see that in the case of a constant coupling we get @xmath51 , which reduces to @xmath52 or roughly @xmath53 we propose potentials that resemble those originally proposed in ref .
@xcite , and in order to be able to explain the coincidence problem we want a potential as simple as possible without any type of cosmological constant .
however all single field potentials previously studied in this context do not fulfill the constraint eq . (
[ eq : cons ] ) .
hence we need a sort of hybrid model as suggested by @xcite . in this type of model
we have effectively two minima .
one is the true minimum which has zero energy density and the other is the false minimum in which our current universe sits .
the offset between the two minima is then what is referred to as the dark energy density .
the dynamics of the scalar field is given by the movement of the false minimum with the dilution of the universe .
one very nice feature of this model is that it is implemented in supersymmetry and hence explains the stability of a very small scalar field mass .
however , since we have an offset it can be interpreted as an effective cosmological constant and as such the hybrid model fails to address the coincidence problem .
in addition , to keep the model stable it is suggested that the lightest neutrino is still relativistic and that we associate dark energy with this .
we address this issue below .
we aim at keeping the model as simple as possible , so we propose the existence of just one scalar field with a democratic coupling to all light neutrino species .
in that case it is assumed that the lightest neutrino does not clump because of pressure , while the two heavier neutrinos become unstable . in the following
we will argue that this is not possible since a feedback from the growth of the heavy neutrino perturbations will cause the relativistic neutrino perturbation to grow as well .
the problem in this scenario can be neatly illustrated by the following equation@xcite @xmath54 the equation explains the growth of the neutrino perturbation and applies to both the interaction between a relativistic neutrino and a scalar field as well as that of a nonrelativistic neutrino and a scalar field .
let us consider a system consisting of three neutrinos , two heavy nonrelativistic neutrinos ( h ) and one light relativistic neutrino ( l ) both interacting with a scalar field .
we only have one scalar field so the nonrelativistic neutrinos will inevitably become unstable towards clumping and @xmath55 grows to very large values . for an interaction with a nonrelativistic neutrino , the average scalar field perturbation can be written as @xcite @xmath56 as we can see the scalar field perturbation is effectively proportional to the neutrino perturbation .
note that in reality there will be extra terms in eq .
( [ averagedphi ] ) since we also have a relativistic neutrino , however , these terms will be sub - dominant compared to the terms from the heavier neutrinos .
so @xmath57 will grow as a result . turning our attention to @xmath58
, we consider the second term in eq .
( [ eq : neutrinop ] ) .
@xmath57 is growing uninhibited to large values , the coupling @xmath32 is for simplicity assumed of order unity ( planck units ) .
@xmath59 is a suppression factor of the order @xmath60 - this factor will act to delay the growth of @xmath58 .
however , since @xmath57 will continue its growth , the inevitable conclusion is that @xmath58 will eventually start to grow .
hence , there exists a type of feedback mechanism between the neutrinos .
note that in this treatment we only considered the interaction between neutrinos of the same type via exchange of the scalar field .
if we include couplings between different neutrinos , the feedback mechanism will be even stronger .
one could of course argue that we are exactly living in a transition regime when @xmath58 has still not turned unstable .
however , that would require serious fine - tuning .
we also investigate a mavan model with a relativistic neutrino numerically using the publicly available cmbfast code @xcite , where we implement the scenario with two heavy neutrinos with assumed masses @xmath61 ev and one species of relativistic neutrinos with assumed mass @xmath62 ev . for the scalar field potential we use a coleman - weinberg@xcite type scalar field potential similar to the one presented in refs . @xcite @xmath63 where @xmath64 and @xmath65 are dimensionful constants , which are fixed by demanding the appropriate amount of dark energy in the universe today as well as solving for the minimum of the effective potential . for the mass term of the three neutrinos we choose the same mass term as in @xcite @xmath66 which results in a coupling @xmath67 .
the code calculates the background energy density and pressure of different species from standard integration methods while at the same time it solves for the minimum of the effective potential of the neutrino - scalar field fluid which is @xmath68 where we remember that there are now two heavy ( h ) neutrinos and one light ( l ) neutrino . the expression above can be recast in a more convenient form as @xcite @xmath69 where we introduced the neutrino number density @xmath70 and neutrino temperature @xmath71 . at the same time as tracking the background
we calculate the perturbations to first order such as @xmath72 this is done for both heavy and light neutrinos using eq .
( [ eq : neutrinop ] ) .
the results are shown in fig .
[ fig : dens ] , where we plot the evolution of the neutrino perturbation and cdm perturbation from a redshift of @xmath73 until today . what happens in fig .
[ fig : dens ] is that the perturbations of the heavy neutrinos , which have already turned nonrelativistic before the range of this plot , are already growing albeit at a slow pace because of the growth - slowing effect of cdm - see eq .
( [ denseom ] ) .
the perturbation of the light neutrino is oscillating ( only visible if zoomed in ) in this epoch , but also growing slightly as a result of the growth of the coupling @xmath32 .
the growth - slowing effect of cdm is , however , only temporary and around a reshift of @xmath74 the heavy neutrino perturbation can no longer resist the temptation towards unstable growth . the heavy neutrino perturbation explodes and because of the increased value of @xmath32 , the light neutrino perturbation starts exploding as well .
we do see that eventually the cdm perturbation also starts exploding and at this moment the linear code is no longer valid .
however at this moment both neutrino perturbations are already in the nonlinear regime and will have started forming neutrino bound states . in this regime
the mavan fluid will no longer act as dark energy and the model breaks down . in conclusion , as a result of a feedback mechanism , the blowing behaviour of the nonrelativistic neutrino density contrast causes the relativistic neutrinos to start clumping as well .
hence the neutrino scalar - field fluid will start acting as a cold dark matter component ( clustering neutrinos ) and hence can not be attributed to dark energy .
and we choose the current neutrino masses @xmath62ev and @xmath61ev ( note that the choice of current neutrino masses does not affect the result qualitatively ) .
we have chosen @xmath75 in cmbfast units of @xmath76 and @xmath77 to fix the cosmology .
the solid black line is cdm - density contrast defined as @xmath78 , the dotted blue line is the heavy neutrino density contrast @xmath79 , and the dashed red line is the light neutrino density contrast @xmath80 .
the heavy neutrinos grow moderately until the coupling becomes large enough for the instabilities to set in .
the light neutrino is still relativistic , and its density contrast oscillates as acoustic waves while growing slightly .
however , due to a feedback mechanism , the relativistic neutrino density contrast tracks that of the nonrelativistic neutrino especially as the growth of the heavy neutrino perturbation becomes unstable ; i.e. both neutrino species will clump . note that the cdm perturbations also blow up at late times .
this is an effect of the system of differential equations breaking down as all parameters go to infinity.,width=264 ] the reason that the relativistic neutrino is able to clump is that it will acquire an effective mass , thus it can not be regarded as a relativistic particle .
unfortunately , we can not use conventional bounds to constrain this effect since the evolution of the neutrino perturbations become nonlinear , i.e. the whole system of equations we are solving , starting with the modified klein - gordon equation breaks down .
this has the effect that all current bounds are no longer valid , as these are established in the linear regime .
single scalar field models can be used to explain late - time acceleration in the mavan scenario . however , in general using these potentials leads to instabilities towards neutrino bound states unless certain criteria are relaxed . in order to obtain a stable or metastable model we direct the reader to ref .
@xcite in which one such model is presented .
the crucial thing is that the scalar field potential resembles that of a cosmological constant . in this case
we simply move the fine - tuning problem to that of explaining the small fraction in the power - law - which unfortunately is also rather unnatural .
accordingly it has been suggested to include an extra scalar field in the treatment .
this has some very nice features and is easily capable of obtaining late - time acceleration as well as @xmath0 today .
however , one drawback is the need for the lightest neutrino to be relativistic today .
as was explained above the feedback mechanism will eventually cause the perturbations of the relativistic neutrino to start growing . as a result , the mavan
fluid will cease to act as dark energy .
we would like to thank lily schrempp , carsten van de bruck , anthony brookfield , david mota , and domenico tocchini - valentini for their great collaboration in the paper establishing the underlying framework for this .
in addition oeb would like to thank neal weiner and subinoy das for some fruitful discussions and nyu for hospitality .
we also acknowledge the use of the publicly available cmbfast code @xcite .
99 c. l. bennett _ et al . _ , astrophys .
j. suppl .
* 148 * ( 2003 ) 1 d. n. spergel _ et al . _ ,
astrophys .
j. suppl .
* 148 * ( 2003 ) 175 d. n. spergel _ et al . _ ,
arxiv : astro - ph/0603449 . m. tegmark _ et al .
_ , phys .
d * 74 * , 123507 ( 2006 ) [ arxiv : astro - ph/0608632 ] .
a. g. riess _ et al . _
[ supernova search team collaboration ] , astron .
j. * 116 * , 1009 ( 1998 ) s. perlmutter _ et al .
_ [ supernova cosmology project collaboration ] , astrophys .
j. * 517 * ( 1999 ) 565 p. astier _ et al .
_ [ the snls collaboration ] , astron .
astrophys . * 447 * , 31 ( 2006 ) w. m. wood - vasey _ et al .
_ , arxiv : astro - ph/0701041 .
p. q. hung , arxiv : hep - ph/0010126 .
p. gu , x. wang and x. zhang , phys . rev .
d * 68 * , 087301 ( 2003 ) r. fardon , a. e. nelson and n. weiner , jcap * 0410 * ( 2004 ) 005 r. d. peccei , phys .
d * 71 * ( 2005 ) 023527 l. schrempp , arxiv : astro - ph/0611912 .
a. ringwald and l. schrempp , jcap * 0610 * , 012 ( 2006 ) d. b. kaplan , a. e. nelson and n. weiner , phys .
lett . * 93 * , 091801 ( 2004 ) v. barger , p. huber and d. marfatia , phys .
lett . * 95 * , 211802 ( 2005 ) m. cirelli , m. c. gonzalez - garcia and c. pena - garay , nucl .
b * 719 * , 219 ( 2005 ) v. barger , d. marfatia and k. whisnant , phys .
d * 73 * , 013005 ( 2006 ) p. h. gu , x. j. bi , b. feng , b. l. young and x. zhang , arxiv : hep - ph/0512076 .
h. li , z. g. dai and x. m. zhang , phys .
d * 71 * , 113003 ( 2005 ) n. afshordi , m. zaldarriaga and k. kohri , phys .
d * 72 * ( 2005 ) 065024 a. w. brookfield , c. van de bruck , d. f. mota and d. tocchini - valentini , phys .
* 96 * ( 2006 ) 061301 a. w. brookfield , c. van de bruck , d. f. mota and d. tocchini - valentini , phys .
d * 73 * ( 2006 ) 083515 r. takahashi and m. tanimoto , jhep * 0605 * ( 2006 ) 021 c. spitzer , arxiv : astro - ph/0606034 .
r. fardon , a. e. nelson and n. weiner , jhep * 0603 * , 042 ( 2006 ) [ arxiv : hep - ph/0507235 ] .
o. e. bjaelde , a. w. brookfield , c. van de bruck , s. hannestad , d. f. mota , l. schrempp and d. tocchini - valentini , jcap * 0801 * ( 2008 ) 026 [ arxiv:0705.2018 [ astro - ph ] ] . m. c. gonzalez - garcia and m. maltoni , phys .
* 460 * ( 2008 ) 1 [ arxiv:0704.1800 [ hep - ph ] ] . k. ichiki and y. y. keum , jcap * 0806 * ( 2008 ) 005 [ arxiv:0705.2134 [ astro - ph ] ] .
p. h. gu , h. j. he and u. sarkar , jcap * 0711 * ( 2007 ) 016 [ arxiv:0705.3736 [ hep - ph ] ] .
n. kaloper , phys .
b * 653 * ( 2007 ) 109 [ arxiv:0706.1977 [ astro - ph ] ] .
l. amendola , m. baldi and c. wetterich , arxiv:0706.3064 [ astro - ph ] . c. wetterich , phys .
b * 655 * ( 2007 ) 201 [ arxiv:0706.4427 [ hep - ph ] ] .
r. bean , e. e. flanagan and m. trodden , new j. phys .
* 10 * ( 2008 ) 033006 [ arxiv:0709.1124 [ astro - ph ] ] .
r. bean , e. e. flanagan and m. trodden , arxiv:0709.1128 [ astro - ph ] .
r. takahashi and m. tanimoto , phys .
d * 77 * ( 2008 ) 045015 [ arxiv:0709.2965 [ hep - ph ] ] .
h. gu , phys .
b * 661 * ( 2008 ) 290 [ arxiv:0710.1044 [ hep - ph ] ] .
f. bauer and l. schrempp , jcap * 0804 * ( 2008 ) 006 [ arxiv:0711.0744 [ astro - ph ] ] .
n. brouzakis , n. tetradis and c. wetterich , arxiv:0711.2226 [ astro - ph ] . j. r. bhatt , p. h. gu , u. sarkar and s. k. singh , phys .
b * 663 * ( 2008 ) 83 [ arxiv:0711.2728 [ hep - ph ] ] .
l. anchordoqui , v. barger , h. goldberg and d. marfatia , phys .
b * 660 * ( 2008 ) 529 [ arxiv:0711.4055 [ hep - ph ] ] . v. pettorino and c. baccigalupi , arxiv:0802.1086 [ astro - ph ] .
d. f. mota , v. pettorino , g. robbers and c. wetterich , phys .
b * 663 * ( 2008 ) 160 [ arxiv:0802.1515 [ astro - ph ] ] .
a. e. bernardini and o. bertolami , phys .
b * 662 * ( 2008 ) 97 [ arxiv:0802.4449 [ hep - ph ] ] .
l. schrempp , p. p. avelino , l. m. g. beca and c. j. martins , phys .
d * 77 * ( 2008 ) 063515 .
k. ichiki and y. y. keum , arxiv:0803.2274 [ hep - ph ] .
k. ichiki and y. y. keum , arxiv:0803.3142 [ astro - ph ] .
j. valiviita , e. majerotto and r. maartens , arxiv:0804.0232 [ astro - ph ] .
j. r. bhatt and u. sarkar , arxiv:0805.2482 [ hep - ph ] .
b. li , d. f. mota and d. j. shaw , arxiv:0805.3428 [ gr - qc ] .
a. goobar , s. hannestad , e. mortsell and h. tu , jcap * 0606 * , 019 ( 2006 ) m. kaplinghat and a. rajaraman , arxiv : astro - ph/0601517 .
u. seljak and m. zaldarriaga , astrophys .
j. * 469 * ( 1996 ) 437 . s. r. coleman and e. weinberg , phys . rev .
d * 7 * , 1888 ( 1973 ) .
r. bean and o. dore , phys .
d * 69 * , 083503 ( 2004 ) s. hannestad , phys .
d * 71 * , 103519 ( 2005 ) c .- p
. ma and e. bertschinger , astrohys . j. * 455 * , 7 ( 1995 ) . | the mass varying neutrino scenario is a model that successfully explains the origin of dark energy while at the same time solves the coincidence problem .
the model is , however , heavily constrained by its stability towards the formation of neutrino bound states when the neutrinos become nonrelativistic .
we discuss these constraints and find that natural , adiabatic , stable models with the right amount of dark energy today do not exist .
secondly , we explain why using the lightest neutrino , which is still relativistic , as an explanation for dark energy does not work because of a feedback mechanism from the heavier neutrinos . |
catalan s constant , so named in honor to eugne c. catalan ( 1814 - 1894 ) , who first developed series and definite integrals representations for it , is a classical mathematical constant which may be defined as @xcite @xmath0 this constant is a special value of some important functions such as the dirichlet s beta function @xmath1 , namely @xmath2 , and the clausen s function @xmath3 , namely denotes the dilogarithm function , defined as @xmath4 for real values of @xmath5 , @xmath6 , and extended to @xmath7 , except for the cut @xmath8 , by analytic continuation . ]
@xmath9 and @xmath10 . for positive integer values of @xmath11
, we can trace an analogy between @xmath12 and @xmath13 , the riemann zeta function , since both @xmath14 and @xmath15 are rational multiples of @xmath16 and @xmath17 , respectively , whereas finite closed - form expressions for both @xmath18 and @xmath19 in terms of other basic constants are unknown @xcite .
however , the proof by apry ( 1978 ) that @xmath20 is irrational @xcite has created an ` asymmetry ' in that analogy because the irrationality of @xmath21 , though very suspected , remains unproven . for which @xmath22 is irrational , and that at least one of the seven numbers @xmath23 is irrational @xcite . ] from the point of view of numerical computation , catalan himself ( 1865 ) computed @xmath24 to @xmath25 decimal places @xcite . by making use of a technique from kummer , bresse ( 1867 )
computed it to @xmath26 decimals , a result that was improved to @xmath27 decimals by glaisher ( 1913 ) @xcite . with the advent of computers , @xmath24 has been computed to a large number of digits .
for instance , yee and chan ( 2009 ) computed it to @xmath28 billion decimals @xcite .
their computation employs two formulas , one of which is a central binomial formula due to ramanujan ( 1915 ) @xcite : , as nicely described in ref . @xcite . ]
@xmath29 on searching for similar rapidly converging series , lupas ( 2000 ) has found the following alternating series @xcite @xmath30 this series converges so fast that it has been implemented in _
_ mathematica__@xmath31 ( version @xmath32 ) for computing @xmath24 . on searching for new congruences modulo primes , z .- w .
sun ( 2011 ) has pointed out that the following central binomial series should converge to @xmath24 ( see conjecture a7 of ref .
@xcite ) : @xmath33 he indeed comments that this formula could be derived from a hypergeometric identity proved by guillera in a recent work @xcite , but a complete proof is not provided there in ref .
@xcite . here in this note ,
starting from a guillera s hypergeometric identity i prove a ramanujan - type series representation for catalan s constant similar to that in eq .
whose convergence rate surpasses that of all known central binomial series representations for this constant .
let us adopt the usual notation for the generalized hypergeometric series : @xmath35 where @xmath36 is the pochhammer symbol .
our main result makes use of the lemma below , which determines a special value for @xmath37 , a function that converges at @xmath38 whenever @xmath39 ( see , e.g. , eq . ( 2.2.1 ) of ref .
@xcite ) .
[ lem:3f2 ] @xmath40 we start from a well - known integral representation for generalized hypergeometric functions ( see , e.g. , eq . ( 1.2 ) of ref @xcite ) , namely @xmath41 valid whenever @xmath42 and @xmath43 .
it then follows that @xmath44 now , let us show that , for all @xmath45 , @xmath46 it is well - known that @xmath47 for all non - null values of @xmath48 for which the hypergeometric series at the left - hand side converges ( see eq .
( 1.5.10 ) of ref .
two sucessive applications of the euler transformation formula @xmath49 on eq .
lead us to @xmath50 and , after some algebra , @xmath51 this completes the proof of eq . .
from eqs . and , one has @xmath52 the trigonometric substitution @xmath53 reduces this integral to @xmath54 which can be solved in terms of the dilogarithm function @xmath55 as follows .
first , note that @xmath56 } + i \left [ \mathrm{li}_2{\left(-e^{i \theta}\right ) } - \mathrm{li}_2{\left(e^{i \theta}\right)}\right ] , \label{eq : primitiva}\ ] ] as can be easily checked by differentiating the right - hand side . then @xmath57 -\left\ { \lim_{a \rightarrow 0^{+}}{a \ , \ln{\!\left [ \frac{1-e^{(i a)}}{1+e^{(i a ) } } \right ] } } + i \left [ \mathrm{li}_2{(-1 ) } - \mathrm{li}_2{(1)}\right]\right\ } \nonumber \\ & = & \frac{\pi}{2 } \ , \ln{\!\left [ \frac{(1-i)^2}{2 } \right ] } + i \left[-2\,i \ , \mathrm{cl}_2{\left(\frac{\pi}{2}\right)}\right ]
-\left\{0 + i \left ( -\frac{\pi^2}{12 } -\frac{\pi^2}{6}\right)\right\ } \nonumber \\ & = & \frac{\pi}{2 } \ ,
\ln{\!\left({-\,i}\right ) } + 2 \ , \mathrm{cl}_2{\left(\frac{\pi}{2}\right ) } + i \ , \frac{\pi^2}{4 } % \nonumber \\ \ : = \ : \frac{\pi}{2 } \ , \left(\ln{1 } -i\,\frac{\pi}{2}\right ) + 2\ , g + i \ , \frac{\pi^2}{4 } \nonumber \\ & = & 2 \ , g \ , , \label{eq:2g}\end{aligned}\ ] ] where the special value of the clausen function in eq .
and the principal value of the logarithm function , with @xmath58 $ ] , were taken into account .
are proved computationally with _ _
mathematica__@xmath31 , one finds the integral formula @xmath59 .
our eqs . and
can then be viewed as a formal proof of this formula . ]
the substitution of this result in eq .
completes the proof .
@xmath60 we are now in a position to prove a rapidly converging central binomial formula for the catalan s constant , which is our main result .
[ teo : main ] @xmath61 let @xmath62}\ ] ] be a function of a real variable @xmath48 in the open interval @xmath63 .
then @xmath64 on the other hand , from guillera s third identity in ref .
@xcite , we know that @xmath65 for all values of @xmath48 for which this series converges .
therefore @xmath66 which implies that @xmath67 on substituting the result obtained in lemma [ lem:3f2 ] for @xmath68 , we find @xmath69 . from eq .
, one then has @xmath70 which completes the proof .
@xmath60 the convergence rate of the just proved central binomial series representation for @xmath71 is to be compared to that of other known similar series , including those in eqs . and .
this is done in details in the next section .
let us now check the convergence rate of each ramanujan - type series for @xmath24 mentioned in this work . by applying the stirling
s improved formula , namely @xmath72 which implies that @xmath73 we shall develop an order estimate of the @xmath74-th term for each series . we begin with ramanujan s series , given in eq . .
its @xmath74-th term is @xmath75 the more complex central binomial series by lupas , see eq .
, has an @xmath74-th term whose absolute value can be estimated as follows : @xmath76 for sufficiently large values of @xmath11 , this simplifies to @xmath77 this convergence rate is clearly slower than that of ramanujan s series , not to say the number of basic arithmetic operations needed to compute the @xmath74-th term , which is considerably larger in the lupas series .
( version @xmath32 ) for computing @xmath34 @xcite . ]
the convergence rate of the series we proved in our theorem [ teo : main ] can be estimated as follows .
the @xmath74-th term of our series is @xmath78 for sufficiently large values of @xmath11 , this can be approximated by @xmath79 the factor @xmath80 makes our series converges faster than both the ramanujan and lupas series . in table
[ tabela ] , below , we compare the error committed in approximating @xmath71 by the partial sum of the first @xmath81 terms corresponding to each central binomial series mentioned in this work .
it is clear from this table that our series is the one that yields the smallest absolute error .
the only competitive series is that conjectured by sun , see our eq . , but a direct comparison of its @xmath74-th term with that of the series in our theorem [ teo : main ] shows that it converges slower . therefore ,
even an eventual proof of sun s conjecture will not furnish a central binomial series faster than the one i am presenting here in this work .
thanks are due to mrs . marcia r. souza and mr .
bruno s. s. lima for checking computationally all hypergeometric series discussed in this work . 19
.deviations from @xmath71 of the partial sums obtained by adding the first @xmath81 terms of each central binomial series mentioned in the text . [ cols="<,<,<,<,<",options="header " , ] | in this note , by making use of a hypergeometric series identity derived by guillera , i prove a ramanujan - type series for the catalan s constant .
the convergence rate of this central binomial series representation surpasses those of all known similar series , including a classical formula by ramanujan and a recent formula by lupas .
interestingly , this suggests that an apry - like irrationality proof could be found for this constant .
catalan s constant , hypergeometric series , central binomial sums , convergence acceleration 11y60 , 11m06 , 30b50 , 33f05 , 33c20 |
the observed @xmath0 and @xmath1 particles are linear combinations of the two mass eigenstates with masses @xmath3 and @xmath4 and a mass difference of @xmath2 . transitions between the two flavor eigenstates are allowed due to non conservation of flavor in weak current interactions and will occur with a frequency proportional to @xmath5 .
together with the mass difference @xmath6 of the @xmath7 system , which has already been measured with high accuracy ( @xmath8@xcite ) , the measurement of @xmath2 is an important ingredient for the precise determination of the side @xmath9 of the ckm unitarity triangle .
the direct determination of @xmath10 and @xmath11 from @xmath12 and @xmath5 is hampered by hadronic uncertainties .
these uncertainties partially cancel in the ratio of mass differences @xmath13 using the experimentally - measured masses @xmath14 and a value for the factor @xmath15 , which can be computed in lattice qcd , the constraint from the ratio @xmath16 is more effective in limiting the position of the apex of the unitarity triangle than the value obtained by @xmath12 measurements alone .
+ @xmath0 oscillations have been observed recently at the fermilab tevatron collider . whereas the d collaboration is reporting a two - sided bound @xmath17 at 90% cl @xcite
, cdf presents the first measurement of the @xmath0-@xmath1 oscillation frequency finding a signal for @xmath18(stat ) @xmath19(sys ) at 95% cl @xcite .
both results are consistent with the prediction of the standard model for the upper bound of @xmath5 @xmath20 @xmath21 @xcite .
+ the work presented here gives an updated estimate for the @xmath2 sensitivity from the atlas experiment using monte carlo events of the @xmath0 hadronic channels @xmath22 and @xmath23 with @xmath24 and @xmath25 .
the following sections contain a brief discussion of the event selection , analysis cuts and the most important kinematic distributions of the @xmath0 candidates .
a detailed description of the generation , simulation , reconstruction and analysis software tools used for this study as well as a short characterization of the properties of the atlas inner detector layout is given in @xcite .
the atlas b - physics trigger with various strategies for b - trigger selections is described in @xcite .
in the offline analysis the @xmath0 meson is reconstructed from its decay products , applying kinematical cuts on tracks , kinematical and mass cuts on intermediate particles like @xmath26 and @xmath27 .
a vertex fit includes mass constraints and requires that the total momentum of the @xmath0 vertex points to the primary vertex and the momentum of the @xmath26 vertex points to the @xmath0 vertex . to improve the purity of the sample cuts on properties of the @xmath0 candidates like proper time , impact parameter , transverse momentum and mass cuts are imposed . for the @xmath22 channel figure [ massbs ]
shows the fitted invariant mass distribution @xmath28 with a mass resolution of @xmath29 = 42.5 mev ( single gauss fit ) .
invariant mass distribution normalized to @xmath30 .
the core of the distribution is fitted with a single gauss function.,width=245 ] the proper time of the reconstructed @xmath0 candidates is computed from the reconstructed transverse decay length @xmath31 , the @xmath0 mass and the @xmath0 transverse momentum @xmath32 .
parameterized with the sum of two gauss functions around the same mean value the widths of the two gaussians resulting from the fit are @xmath33 = ( [email protected] ) fs for the core fraction of 54.7% and @xmath35 = ( [email protected] ) fs for the rest of the tail part of the distribution . for @xmath0 from @xmath36
decays and fitted with the sum of two gauss functions around a common mean value.,width=245 ] the estimation of the maximum value of @xmath5 measurable with atlas is considering @xmath0 candidates from the @xmath37 and @xmath38 channels .
the detailed analysis is done for the @xmath22 channel , whereas numbers for the @xmath23 signal and @xmath39 exclusive background channels are estimated extrapolations .
more details on the event selection including cut values , kinematical resolutions , expected number of events for signal and background channels can be found in @xcite .
the probability density to observe an initial @xmath40 meson ( @xmath41 ) decaying at time @xmath42 after its creation as a @xmath43 meson is given by @xmath44 where @xmath45 , @xmath46 and @xmath47 .
+ for the unmixed case ( an initial @xmath40 meson decaying as a @xmath40 meson at time @xmath42 ) , the probability density is obtained by setting @xmath48 in eq .
[ oscpr ] .
+ experimental effects like the wrong tag fraction and the resolution of the reconstructed proper time of the @xmath0 modify the probability function .
exclusive oscillating background channels and non - oscillating combinatorial background are taken into account as fractions of the probability density called pdf@xmath49 .
+ the likelihood of the total sample is written as @xmath50 the index @xmath51 denotes the @xmath36 channel and @xmath52 the @xmath53 channel , @xmath54 is the total number of events of type @xmath55 , and @xmath56 .
see @xcite for detailed information on the building of the probability density functions .
a simplified monte carlo method is applied to generate a @xmath0 sample using the numbers of reconstructed @xmath0 events and kinematic distributions obtained from the simulation studies in ref . @xcite as input parameters .
@xmath0 signal events oscillating with a given frequency @xmath5(e.g .
@xmath57 , which is far off the expected value for @xmath5 ) , together with @xmath58 background events oscillating with frequency @xmath12 and @xmath59 combinatorial events ( no oscillations ) are generated according to eq .
[ oscpr ] .
the @xmath5 measurement limits are obtained applying the amplitude fit method @xcite to the ` data sample ' generated as described in the previous section . according to this method a new parameter , the @xmath0 oscillation amplitude @xmath60 ,
is introduced in the likelihood function by replacing the term ` @xmath61 ' with ` @xmath62 ' in the @xmath0 probability density function given in eq .
[ oscpr ] . for each value of @xmath5 ,
the new likelihood function is minimized with respect to @xmath60 , keeping all other parameters fixed , and a value @xmath63 is obtained .
one expects , within the estimated uncertainty , @xmath64 for @xmath5 close to its true value , and @xmath65 for @xmath5 far from the true value .
a @xmath66 measurement limit is defined as the value of @xmath5 for which @xmath67 , and a sensitivity at 95% confidence level as the value of @xmath5 for which @xmath68 .
limits are computed with the statistical uncertainty @xmath69 . a detailed investigation on the systematic uncertainties @xmath70 , which affects the measurement of the @xmath0 oscillation ,
is presented in @xcite .
+ for the nominal set of parameters ( as defined in the previous sections ) , @xmath71 and an integrated luminosity of 30 @xmath72 the amplitude @xmath73 is plotted as a function of @xmath5 in fig .
[ ampl_dms ] . the 95% cl sensitivity to measure @xmath5
is found to be 30.5 @xmath74 .
this value is given by the intersection of the dashed line , corresponding to @xmath75 with the @xmath64 horizontal line .
+ oscillation amplitude as a function of @xmath2 for an integrated luminosity of 30 @xmath72 for a specific monte carlo experiment.,width=226 ] from fig .
[ sign_dms ] , which shows the significance of the measurement @xmath76 as a function of @xmath5 , the @xmath77 measurement limit is found to be 22 @xmath74 . for an integrated luminosity of @xmath78.,width=226 ] .the dependence of @xmath2 measurement limits on the integrated luminosity . [ cols="^,^,^ " , ] the dependence of the @xmath5 measurement limits on the integrated luminosity is shown in fig .
[ dms_lum ] , with the numerical values given in table [ tab_dms_lum ] .
measurement limits on the integrated luminosity .
the dotted horizontal line gives the cdf measurement @xcite.,width=245 ] the dependence of the @xmath5 measurement limits on @xmath79 is determined for an integrated luminosity of 30 @xmath72 , other parameters having their nominal value .
the @xmath79 is used as a fixed parameter in the amplitude fit method . as shown in fig .
[ dms_dg ] no sizeable effect is seen up to a @xmath80 of 50% .
+ measurement limits on @xmath79.,width=245 ]
in this summary the performance of the @xmath81 channel and extrapolated numbers for @xmath23 channel are used to calculate the 95% cl exclusion and @xmath66 measurement limits of the @xmath82 oscillation frequency as a function of the integrated luminosity collected with the atlas detector .
the limits are updated for the detector geometry of `` initial layout '' using full rome statistics , but only statistical errors are taken into account . with an integrated luminosity from 10 to 20 @xmath72
a @xmath66 measurement for a range of @xmath83 is possible , covering the recent results from the tevatron collider .
+ the values obtained in this note for the measurement limits should be re - evaluated , taking into account changes in the detector geometry , especially `` complete detector layout '' , and the evolving simulation and reconstruction software . @xmath38 and
exclusive @xmath39background channels will be analyzed independently and investigations looking at the performance of other interesting @xmath0-@xmath1 mixing channels , which might be included in the analysis , will be carried out . 9 s. eidelman et al .
[ particle data group ] , `` review of particle physics '' , phys .
b592 , 1 , 2004 .
d0 collaboration , `` first direct two - sided bound on the @xmath0 oscillation frequency '' , physical review letters 97 , 021802 , 2006 .
cdf collaboration , `` measurement of the @xmath0 - @xmath1 oscillation frequency '' , physical review letters 97 , 062003 , 2006 . m. battaglia , a. buras , p. gambino and a. stocchi , `` workshop on the ckm matrix and the unitarity triangle '' , hep - ph/0304132 and cern-2003 - 002 , 2003 .
b. epp , e. kneringer , h. duer , p. jussel , `` performance of the @xmath84 channel with the atlas detector and prospects for the measurement of @xmath82 oscillations using rome mc simulation '' , atl - phys - pub-2006 - 020 , 2006 . | an estimation of the sensitivity to measure @xmath0-@xmath1 oscillations with the atlas detector is given for the detector geometry of `` initial layout '' .
the @xmath2 reach is derived from unbinned maximum likelihood amplitude fits using @xmath0 events generated with a simplified monte carlo method . |
the study of superconducting fluctuations above the superconducting transition temperature , @xmath3 , in iron - based superconductors @xcite has gain increasing attention @xcite , as these effects extend up to relatively large temperatures above @xmath3 , when compared with low-@xmath3 superconductors.@xcite these enhanced fluctuations as observed in pnictides are due to their layered structure in conjunction with their relatively large @xmath3 , large values of the ginzburg - landau ( g - l ) parameter , @xmath9 , and large values of the upper critical field , @xmath10 ( low values of the coherence length ) . studies of superconducting fluctuations , near and above @xmath3 , allow to compare experiments with available theoretical predictions for critical and gaussian fluctuations in superconductors .
as fluctuations are sensitive to the pairing symmetry , it is of particular interest to study pnictides , for which multi - band ( inter - band ) levels participate in the pairing @xcite producing a @xmath11 symmetry ( the symmetry is still not clear , and seems to depend on the doping @xcite ) .
for instance , a recent theoretical work taking into account the characteristic inter - band pairing of the pnictides and a @xmath11 symmetry @xcite predicted that conductivity fluctuations should exhibit only one critical regime , while two critical regimes are expected near @xmath3 for a s - wave pairing ( a statical one , followed by a dynamical).@xcite superconducting fluctuations can also provide information on the system dimensionality .
analysis of critical and gaussian fluctuations in the studied pnictides in the literature appears to support a three dimensional , 3d , behavior @xcite but two dimensional behavior was observed for higher fields ( above 8 t ) in smfeas system @xcite and in lifeas @xcite , and a recent theory of critical fluctuations was specially developed for pnictides considering an intermediated dimensionality between two and three for high applied magnetic fields which seems to explain high - field experimental data as well.@xcite the ginzburg criterium which defines the temperature width of critical fluctuations close to @xmath3 can be estimated by the relation @xmath12
@xmath13 1.07x@xmath14 k @xcite where @xmath10 is the field obtained by extrapolating the linear region near @xmath3 in the @xmath15 vs. @xmath5 diagram to @xmath5 = 0 k. values for conventional superconductors are @xcite @xmath12 @xmath13 @xmath16 k ,
@xmath12 @xmath17 1 k for high-@xmath3 cuprates @xcite , but of the order of @xmath18 k for pnictides.@xcite this predicted narrow critical region , difficult to be resolved experimentally , can increase orders of magnitude under the effect of a high - magnetic field that constricts the pairs to the lowest landau level , lowering the system dimensionality.@xcite the latter explains the considerably large critical fluctuation regime that has been observed in pnictides under high applied magnetic fields.@xcite in most of the cases the study of high - field critical fluctuations is possible by checking if experimental curves follow specific scaling laws derived from the g - l theory developed for layered systems within the lowest - landau - level approximation , lll , and taking in to account interactions in the fluctuation regime.@xcite these scalings laws when applied to experimental data may allow to estimate many values of g - l intrinsic parameters @xcite , but unfortunately , do not provide even a qualitative information regarding the temperature width where these fluctuations are important .
conductivity fluctuations above the surperconducting transition temperature , also known as excess conductivity or paraconductivity , have been widely theoretically and experimentally studied in most of known superconductors @xcite , including earlier works studying gaussian fluctuations at zero magnetic field within the well known aslamazov - larkin , a - l , theory.@xcite in more recent experimental works performed in the presence of magnetic fields , fluctuation effects are interpreted in terms of extensions of the a - l theory @xcite when in the gaussian regime , and , near the transition temperature , by scaling - laws derived from g - l theories which include correlations in the fluctuation regime , then critical.@xcite for pnictides , magneto - conductivity studies were performed in smfeaso @xcite , fluor doped smfeaso @xcite , ( nd , pr)feaso @xcite , ( nd , pr , sm)feaso @xcite , bafe@xmath19co@xmath20as@xmath2 @xcite , ba@xmath7k@xmath20fe@xmath2as@xmath2 @xcite , laofeas @xcite , lifeas @xcite and in bafe@xmath19ni@xmath8as@xmath2 @xcite . for most of these studies
@xcite data were analyzed in terms of scaling - law approaches for critical fluctuations , while data in refs .
@xcite , @xcite and @xcite were analysed in terms of a 3d a - l theory @xcite or its extensions.@xcite in this work we study isofield resistivity curves previously obtained @xcite in a high - quality bafe@xmath0ni@xmath1as@xmath2 single crystal @xcite with @xmath3 = 20.0 k and @xmath21@xmath3 @xmath22 0.3 k ( electron - doped 122 system ) .
we focus our analysis on the width of the critical regime above the superconducting transition , the associated critical exponents , and the evolution of this regime with the applied magnetic field .
as will be shown , the results resolved for the first time the width of the critical region under applied magnetic field and its associated critical exponent .
as detailed in ref .
@xcite , precise isofield resistivity , @xmath23 , measurements were obtained as a function of temperature , for magnetic fields running from zero to 9 t applied parallel to the c - axis of the sample .
the excess conductivity , @xmath4 , above the superconducting temperature transition is obtained from each isofield resistivity curve by subtracting the normal state resistivity , @xmath23@xmath24 : @xmath4=1/@xmath23 - 1/@xmath23@xmath24 .
the normal state resistivity is obtained by fitting the form @xmath23@xmath24=b+c@xmath5+d@xmath5@xmath25 in a selected temperature region ( within a temperature width of 30 k ) in the normal state .
figure 1 displays the resulting conductivity curves .
the upper inset of fig .
1 shows a selected resistivity curve and the corresponding fitting performed in the normal state region extrapolated to lower temperatures .
the lower inset shows d@xmath23/d@xmath5 as obtained from the resistivity curve with @xmath15=0 , evidencing the sharp transition .
the solid line crossing the isofield conductivity curves in fig.1 represents a virtual line separating critical and gaussian fluctuation regimes which discussion is presented below .
arrows in each curve of fig.1 indicate values of @xmath6 obtained from the lll scaling , which is also discussed below .
( @xmath15 ) vs. t curves .
the upper inset shows a selected resistivity curve and the fitting performed on the normal state region , @xmath23@xmath24 , extrapolated to lower temperatures .
the lower inset shows d@xmath23/d@xmath5 for @xmath15 = 0 . ]
from the aslamazov - larkin theory @xcite , the excess conductivity , @xmath4 , above @xmath3 for a three dimensional superconductor is proportional to the coherence length @xmath26 .
then , the excess conductivity above @xmath3 is expected to behave as @xcite @xmath27 where a is a constant and @xmath28 an exponent which assumes different values depending on the regime of fluctuations .
this equation is expected to hold in the presence of a magnetic field by just replacing @xmath3 by @xmath6 . in the absence of magnetic fields ,
the expected values of the exponent @xmath28 are : @xmath28=1/2 in the mean field regime , crossing over to @[email protected] as temperature approaches @xmath3 corresponding to a static critical exponent , crossing over to @[email protected] even closer to @xmath3 corresponding to a dynamical critical exponent @xcite .
the equation above can also be written as -1/(dln(@xmath4)/d@xmath5)=(1/@xmath28)(@xmath5-@xmath3 ) , and analysis of experimental conductivity curves in terms of plots of -1/(dln@xmath4/d@xmath5 ) vs. @xmath5 may allow to estimate exponent values and temperature widths in the fluctuation regime.@xcite here , we apply this approach , among others , to analyze the conductivity curves presented in fig.1 .
figure 2a and 2b show plots of the quantity -1/(dln(@xmath4)/d@xmath5 ) vs. @xmath5 as obtained from the conductivity curve for @xmath15 = 0 and @xmath15 = 1 t respectively .
the derivatives dln(@xmath4)/d@xmath5 were calculated by using the commercial program origin 8 . as shown in fig .
2a , it is possible to identify two distinct linear regions above @[email protected] k. the first region with a width @xmath21@[email protected] k has an exponent @[email protected] , which is close to the mean field value @xmath28=0.5 and it appears to be associated to three dimensional a - l fluctuations , as discussed above .
an extrapolation of this region to the @xmath5 axis , sugests the value of @xmath3 = 20.2 k. the region below 20.2 k in fig .
2a is likely to be associated with the transition width @xmath220.3k .
as expected , the experiment can not resolve any critical region near @xmath3 , which would occur within a 10@xmath29 k width .
a second region with @xmath28@xmath223 starts at @xmath5@xmath22 21.1 k and seems to extend up to higher temperatures although a large spread of the data is observed above 23 - 24 k ( not shown ) .
the inset of fig .
2a shows the original ln@xmath4 vs. @xmath5 curve of the main figure and the dashed line is the fitting of the data following the 3d a - l theory presented in ref.@xcite .
arrows in this inset give a rough indication of the exponents @xmath28 in the curve .
between this two regions , the plot in fig .
2a shows a small region with decreasing curvature from which , one can barely resolve an exponent @[email protected] .
probably this small region has no associated exponent and just represents a decay in the amplitude of a - l fluctuations , which is physically expected to occur as temperature increases above @xmath3 ( entering in the region with @xmath28@xmath223 ) .
such decay in the amplitude fluctuations are related to the well - known short - wave length effects occurring well above @xmath3 @xcite , and explains the rather large value of the exponent @xmath28@xmath223 associated with a - l gaussian fluctuations in the corresponding region.@xcite figure 2b shows the same plot of fig .
2a , but with a magnetic field @xmath15 = 1 t , which exhibits a totally different scenario .
it is possible to resolve three different linear regions in this figure .
the first region , with @[email protected] starts just above the irreversible temperature , @xmath30 ( @xmath30 is defined here as the temperature below which a critical current @xmath31@xmath320 is established ) , and ends close to the mean field transition temperature , @xmath6 , which is estimated by extrapolating the second regime with @[email protected] to the @xmath5-axis ( marked with an arrow ) . above @xmath6
there are two distinct regions , the first , with @[email protected] , is most likely to be related to the mean field a - l fluctuations under a magnetic field ( similar to the a - l fluctuations with @[email protected] observed in fig .
2a with @xmath15 = 0 ) , and the second one , with @[email protected] , is likely to be associated to a - l fluctuations with short - wave length effects since data in this region lies in the same region where the amplitude fluctuations decay to zero as temperature increases.@xcite ) /d@xmath5 is plotted against @xmath5 for : a ) @xmath15 = 0 ; and b ) @xmath15 = 1 t. the inset in a ) shows ln(@xmath4)vs.@xmath5 for @xmath15=0 and a 3d - al theory fitting extracted from ref.@xcite ; the inset in b ) shows @xmath33@xmath34 vs. @xmath15 . ] ) /dt is plotted against t for h@xmath353 t .
vertical arrows up show the position of @xmath6 , obtained from the 3d - lll scaling , in each respective curve . ] / @xmath36mk@xmath37 ) ] figure 3 , shows plots of -1/dln(@xmath4)/d@xmath5 vs @xmath5 for magnetic fields running from 3 t to 9 t. it is interesting to observe in this figure that all curves exhibit only two different regions , each one with approximately the same exponent ( the exponent of the region starting at @xmath30 varies between 2.1 for @xmath15 = 3 t to 2.6 for @xmath15 = 9 t , similar to the region starting at @xmath30 for @xmath15 = 1 t in fig .
2b , and the exponent of the second region varies between 1.3 and 1.4 similar to the second region above @xmath6 in fig .
2b ) , which appears to be a clear effect of the high magnetic field .
the absence of a linear region occurring between these two regions , as observed for @xmath15 = 1 t in fig .
2b with @[email protected] , impeded us to estimate a value for @xmath6 in each curve of fig .
3 . the second region in the curves of fig .
3 with @[email protected] are likely to be associated with a - l fluctuations with short - wave length effects ( in analogy to the region with @[email protected] in fig .
2b).@xcite the first linear regions starting at @xmath30 in the curves of fig . 3 ( and of fig .
2b ) with exponents @xmath28 varying between 2.1 to 2.6 corresponds to the reversible ( flux - flow ) regions observed in the resisitivity curves which show a rounding effect ( as resisitivity approaches the normal region ) due to thermal fluctuations .
many types of fluctuations have been proposed to explain the flux - flow region in type ii superconductors , including glass - like fluctuations occurring near ( above ) the glass transition temperature @xmath38 ( the glass phase which shows zero resistivity below @xmath39@xmath22@xmath30 may appear from a continuos transition that takes place in the vortex - liquid phase , driven by disorder ) @xcite , low - field 3d - xy fluctuations ( driven by phase fluctuations of the complex order parameter as in superfluid @xmath40he ) @xcite and high field lll fluctuations @xcite . in the case of lll fluctuations ,
the high magnetic field constrict the pairs to the lowest - landau - level which acquires an effectively 1d dimension enhancing the importance of fluctuations . as it is shown in ref.@xcite the inclusion of the quartic @xmath41 term in the g - l hamiltonian within the lll - aproximation
( this term is only important in the critical region near @xmath6 ) removes the divergence at the mean field @xmath6 and as a consequence the mean field @xmath42 line becomes a crossover line . since fluctuations of 3dxy type have been used to explain the vortex - liquid phase in resisitivity and magnetization curves in the low field regime @xcite , we applied a 3dxy scaling for conductivity to our isofield @xmath43 curves of fig .
1 , where a plot of log@xmath44(@xmath43@xmath15@xmath45 vs. ( 1-@xmath46)@xmath15@xmath47 ( where @xmath3 is the only adjustable parameter , and 1/@xmath48=-0.747 where @xmath49@xmath50=-0.669 ) are expected to collapse in to a single curve . figure 4b show this plot with the parameter @xmath3 = 20 k , evidencing that the lower field curve for @xmath15 = 1 t fail to follow the scaling , which excludes this type of fluctuations . for the case of glass - like fluctuations occurring near ( above ) @xmath38 @xcite , the conductivity is expected to follow @xmath51 with the critical exponent @xmath28= @xmath49(2+@xmath52-@xmath53 ) where @xmath49 is the coherence lenght exponent , @xmath52 is the dynamical critical exponent and @xmath53 is the dimensionality .
this type of fluctuations have been considered in the 3d `` gauge glass '' model ( in the vortex representation ) presented in ref.@xcite where it is obtained @xmath49=1.3 and @[email protected] producing @[email protected] ( @xmath53=3 ) , which value is close to the values of the exponent @xmath28 found in figs . 3 and 2b .
this agreement suggests that the first linear regions of fig .
2b and 3 can be explained in terms of gauge - glass fluctuations @xcite occurring around ( above ) @xmath30 . in that case , a second order phase transition ( possibly a glass phase ) should take place at some temperature very close to @xmath30 . since the glass transition is driven by disorder , it is important to compare the irreversibility line obtained from @xmath30 values extracted from the curves of figs .
2b and 3 with the expression developed in ref.@xcite for the ( glass line ) irreversibility line which considers the effect of disorder : 1-@xmath54-@xmath55 + 2[@xmath56(1-@xmath54)@xmath25@xmath55/4@xmath57@xmath37[3/2-(4@xmath58@xmath54@xmath59/(@xmath56(1-@xmath54)@xmath25 ) ] = 0 , where @xmath54=@xmath5/@xmath3 ( we used @xmath3=19.83 k ) , @xmath55=@xmath15/@xmath10 ( we used @xmath10 = 42 t ) , and @xmath56 and @xmath60 ( which is related to the ginzburg number ) are fitting parameters defined in ref@xcite representing the disorder and the strength of thermal fluctuations respectively ( for 1ma , the electric field in the sample is @[email protected] v / m , allowing us to use the static form of the irreversibility line expression in ref.@xcite ) . figure 4 shows a ( @xmath15,@xmath5 ) diagram for the irreversibility line where the solid line represents a best fitting to the above expression obtained with @xmath56=0.003 and @xmath60=1x10@xmath62 ( this value of @xmath60 , is two orders of magnitude larger than the value encountered in ref.@xcite for nbse@xmath2 ) .
since our value of @xmath56 is larger than the value encountered for nbse@xmath2 in ref.@xcite it suggests that disorder is important in the studied sample , supporting the hipothesis of glass - like fluctuations occurring near ( above ) @xmath30 .
despite we did not obtain @xmath63 magnetization curves to check for the possible existence of a first order melting line and of the peak effect , we mention that @xmath63 curves obtained in a similar sample of bafe@xmath0ni@xmath1as@xmath2 presented in ref.@xcite could not resolve any discontinuity in magnetization in the vicinity of @xmath64 and also show that the second magnetization peak , centered at @xmath65 , observed in @xmath63 curves dissapears as temperature approaches @xmath3 with the corresponding @xmath65 line ending at an isolated point in the ( h , t ) phase diagram , below the irreversiblity line
. it would be interesting to compare values of the exponent @xmath28 obtained here with those obtained for other superconductors , but unfortunatelly , there is an apparent lack of similar studies in the literature . in any case
, the observation of only one possible critical region appears to be in agreement with the predictions of ref .
@xcite for pnictides .
it is interesting to analyse the behavior of the width , @xmath33@xmath34 , with @xmath15 of the first linear region ( with @xmath28@xmath322 ) appearing in the curves of fig . 3 and of fig .
2b . a plot of the quantity @xmath33@xmath34 vs. @xmath15 is shown in the inset of fig . 2b , and the best fit of the resulting curve produced @xmath33@xmath5@xmath66@xmath67 which is in direct agreement with the field - dependent ginzburg criterium , @xmath21@xmath34@xmath66@xmath68 , predicted for three dimensional fluctuations of lll type.@xcite to check this possibility
, we apply the 3d - lll scaling law derived in ref .
@xcite for magneto - conductivity , to the isofield curves of fig .
1 with @xmath15 running from 1 to 9 t. figure 4c shows the results of this scaling applied on data lying in the first linear region of fig . 3 and of fig .
2b , where @xmath69 is plotted against ( @xmath5-@xmath6)/(@xmath70)@xmath37 and @xmath6 ( representing the mean field transition temperature ) is an adjustable parameter .
the collapse of the curves suggests that lll - fluctuations also play a role in the first linear regions of the curves of figs .
it should be mentioned that calculations performed in ref.@xcite show that the lll contribution to conductivity dominates above the glass transition ( flux - flow regime ) where @xmath71@xmath72 is a function of @xmath73 and the lll scaled temperature a function of ( @xmath74)@xmath75 , which suggests a scaling law for conductivity curves in the vortex - liquid phase similar to the 3d - lll scaling law used .
the lll scaling procedure allowed to obtain values of the mean field temperature transition @xmath6 producing a reasonable value for @xmath76d@xmath77/d@xmath5 = -3 t / k .
the values of @xmath6 are plotted with values of @xmath30 in fig .
4a . to better visualize the results of this work
, we added an arrow in each curve of fig . 3 and of fig .
1 indicating the position of @xmath6 obtained from the scaling approach , and a line linking the values of conductivity at the virtual points separating critical and gaussian regimes in fig .
3 ( dotted line ) and in fig .
1 ( solid line named @xmath78 line ) .
in conclusion , analysis of high - field conductivity curves evidence the existence of only two fluctuation regimes , a possible critical one going from the irreversible temperature to above @xmath6 , followed by aslamazov - larkin fluctuations in the gaussian regime .
our analysis suggests that critical fluctuations emerge above @xmath30 under high applied magnetic fields .
these critical fluctuations , not resolved for zero magnetic field , appear to account for the excess conductivity above @xmath6 , which extends more than 1 k for 9 t. the analysis allowed to estimate with certain accuracy the width of the critical region as well as its associated critical exponent .
work supported by : cnpq and faperj ( brazil ) ; the spanish micinn and erdf ( no .
fis2010 - 19807 ) , and the xunta de galicia ( nos .
2010/xa043 and 10tmt206012pr ) ( spain ) ; nsfc program ( no .
11004233 ) and 973 project ( no .
2011cba00110 ) ( p.r .
we thank discussions with s.l.a . de queiroz and p. pureur
. 10 kamihara y , watanabe t , hitano m and hosono h 2008 _ j. am . chem .
soc . _ * 130 * 3296 rey r i , carballeira c , mosqueira j , salem - sugui jr .
s , alvarenga a d , luo h - q , lu x -y , chen y - c and vidal f 2013 _ supercond .
technol . _
* 26 * 055004 pallechi i , faciulli c , tropeano m , palenzona a , ferretti m , malagoli a , martinelli a , sheikin i , putti m and ferdeghini c 2009 _ phys .
b _ * 79 * 104515 salem - sugui jr .
s , ghivelder l , alvarenga a d , pimentel jr .
j l , luo h - q , wang z and wen h - h 2009 _ phys .
b _ * 80 * , 014518 mosqueira j , dancausa j d , vidal f , salem - sugui jr .
s , alvarenga a d , luo h - q , wang z .- s and wen h .-
h 2011 _ phys .
* 094519 ; ibid . 2011 _ phys . rev .
b _ * 83 * , 219906(e ) choi c , hyun kim s , choi k - y , jung m - h , lee s .- i , wang x f , chen x h , and wang x l 2009 _ supercond .
technol . _
* 22 * 105016 kim s h , choi c h , jung m - h , yoon j - b , jo y - h , wang x f , chen x h , wang x l , lee s - k , and choi k - y 2010 _ j. appl .
phys . _ * 108 * 063916 fanfarillo l , benfatto l , caprara s , castellani c , and grilli m 2009 _ phys .
b _ * 79 * 172508 muray j m and tesanovic z 2010 _ phys .
lett . _ * 105 * 037006 liu s l , longyan g , gang b , haiyun w and yongtao l 2011 _ supercond .
technol . _
* 075005 liu s l , gang b , and haiyun w 2010 _ j. supercond .
nov magn _ * 23 * 1563 liu s l , haiyun w and gang b 2010 _ phys . lett .
a _ * 374 * 3529 liu s l , li y , bao g , and wang h 2011 _ j. supercond .
nov magn _ * 24
* 1971 liu s l , haiyun w and gang b 2011 _ sol .
state comm . _
* 151 * 1 pandya s , sherif s , chandra l s s and ganesan v 2011 _ supercond .
technol . _ * 24 * 045011 pandya s , sherif s , chandra l s s and ganesan v 2010 _ supercond .
technol . _
* 23 * 075015 solovjov a l , sidorov s l , tarenkov y v and dyachenko a i 2009 _ low temp . phys . _ * 35 * 826 prando g , lascialfari a , rigamonti a , romano l , sanna s , putti m and tropeano m 2011 _ phys .
b _ * 84 * 064507 rullier - albenque f , colson d , forget a and alloul h 2012 _ phys .
lett . _ * 109 * 187005 mosqueira j , dancausa j d , carballeira c , salem - sugui jr .
s , alvarenga a d , luo h - q , wang z - s , wen h - h and vidal f 2013 _ j supercond .
nov magn _ * 26 * 1217 tinkham m 1996 _ introduction to superconductivity _
( mcgraw - hill , new york ) cvetkovic v and tesanovic z 2009 _ europhys .
* 85 * 37002 zehetmayer m 2013 _ supercond .
technol . _ * 26 * 04300 lobb c j 1987 _ phys .
b _ * 36 * 3930 lee w c , klemm r a and johnston d c 1989 _ phys .
lett . _ * 63 * 1012 pureur p , menegotto costa r , rodrigues jr .
p , schaf j and kunzler j v 1993 _ phys .
b _ * 47 * 11420 salem - sugui jr .
s , alvarenga a d , vieira v n and schilling o f 2006 _ phys .
b _ * 73 * 012509 rosenstein b and li d 2010 _ rev .
phys . _ * 82 * 109 ullah s and dorsey a t 1990 _ phys .
lett . _ * 65 * 2066 ; ibid .
b _ * 44 * , 262 rosenstein b , shapiro b ya , prozorov r , shaulov a and yeshurun y 2001 _ phys . rev .
b _ * 63 * 134501 aslamazov l g and larkin a i 1968 _ phys . lett .
* 26 * 238 carballeira c , currs s r , via j , veira j a , ramallo m v and vidal f 2001 _ phys . rev .
b _ * 63 * 144515 chen y , lu x , wang m , luo h - q and li s 2011 _ supercond .
technol . _
* 24 * 065004 johnson w e , tsuei c c and chaudhari p 1978 _ phys .
b _ * 17 * 2884 vidal f , carballeira c , currs s r , mosqueira j , ramallo m v , veira j a and via j 2002 _ europhys . lett . _ * 59 * 754 fisher m p a and huse d a 1991 _ phys . rev .
b _ * 43 * 130 salamon m b , shi j , overend n and howson m a 1993 _ phys . rev .
b _ * 47 * 5520 friesen m and muzikar p 1998 _ europhys .
* 43 * 65 wengel c and young p a 1997 _ phys .
b _ * 56 * 5918 rosenstein b and zhuravlev v 2007 _ phys . rev .
* 014507 salem - sugui jr s , ghivelder l , alvarenga a d , cohen l f , luo h - q and lu x 2013 _ supercond sci . technol . _ * 26 * 025006 | we study fluctuation conductivity in a single crystal of bafe@xmath0ni@xmath1as@xmath2 superconductor ( @xmath3 = 20 k ) as a function of temperature and applied magnetic field .
magneto - conductivity curves , @xmath4 vs. @xmath5 , were analyzed in terms of -1/(dln(@xmath4)/d@xmath5 ) vs. @xmath5 plots , which allow to study different fluctuations regimes and to estimate exponent values and temperature widths of each regime .
the analysis of magneto - conductivity curves evidence the existence of only two fluctuations regimes , a possible critical one ( of glass - like type ) going from the irreversible temperature to above @xmath6 followed by aslamazov - larkin fluctuations in the gaussian regime .
keywords : bafe@xmath7ni@xmath8as@xmath2 , magneto - conductivity , critical exponents |
this work was supported by the u.s . army research office ( w911nf-13 - 1 - 0024,w911nf-12 - 0607 ) , the australian research council ( ce11e0001017 ) , and the nsw node of australian national fabrication facility .
computational resources on nanohub.org , funded by the nsf grant eec-0228390 , were used . | silicon ( si ) quantum dots ( qd ) have been among the most prominent candidates for implementing spin based qubits with a potential for scalability , due to their exceptional coherence times . to build a large - scale quantum computer with si qds
, we must address any dot - to - dot variations that can cause randomness in qubit operations . in this work
, we identify the presence of monoatomic steps at the si / sige or si / sio@xmath0 interface as a dominant source of variations in the electron @xmath1-factors , which will cause qubit - to - qubit variability in the single qubit spin resonance frequencies . using atomistc tight - binding calculations
we show that the stark shift of the resonance frequencies , essential to selectively address individual spin qubits , also undergoes variations due to these steps .
we compare our theoretical predictions with experiments on qds at a si / sio@xmath0 interface , in which we observe significant differences in stark shifts between qds in two different samples .
we also experimentally observe variations in the @xmath1-factors of one - electron and three - electron spin qubits realized in three neighboring qds on the same sample , at a level consistent with our calculations .
the dephasing times of these qubits also vary , most likely due to their varying sensitivity to charge noise .
more importantly , from our calculations we show that with the help of the anisotropic nature of the spin - orbit interaction ( soi ) in a si qd , we can minimize and control these variations .
ultimately , we predict an increase in the dephasing time of the si qd spin qubits by at least an order of magnitude , by aligning the external dc magnetic field towards specific crystal orientations .
a scalable quantum computing architecture requires reproducibility and control over key qubit properties , such as resonance frequency , coherence time , etc .
variability in such parameters among qubits of a large - scale quantum computer would necessitate individual characterization and control@xcite of every qubit in that computer , while excessive variability could even make scaling impractical . in order to assess and possibly mitigate these issues we need to identify and understand potential sources of device - to - device variability in critical qubit behavior .
spin qubits hosted in silicon ( si ) quantum dots ( qd)@xcite have been showing promise as a potential building block for a large - scale quantum computer@xcite , because of their compatibility with already existing cmos technology and the long coherence times available due to the presence of negligible nuclear spins in isotopically purified @xmath2si@xcite . single@xcite and
two qubit@xcite gates have been demonstrated already . to move forward with increasing number of qubits@xcite
, we have to study possible sources of variability and randomness in these qubits .
the energy levels of interest in a si qd for qubit operations are two low lying valley states @xmath3 and @xmath4 , each split in two spin levels in the presence of a dc magnetic field , @xmath5 . until recently
, the main focus has been to improve the valley splitting energy to create well isolated states . in metal - oxide - semiconductor
based si qds , this is now routinely achieved@xcite .
however , it turns out that the @xmath1-factor and thus the spin splitting ( @xmath6 , where @xmath7 is the bohr magneton ) is valley dependent@xcite and as we will show experimentally , is sample - to - sample dependent . in a si quantum well or dot , presence of structure inversion asymmetry ( sia ) introduces the rashba spin - orbit interaction ( soi)@xcite . though it is known that due to the lack of bulk inversion asymmetry ( bia ) , the dresselhaus soi is absent from bulk si , interface inversion asymmetry ( iia ) contributes a dresselhaus - like term in interface confined structures in si@xcite . both the rashba and dresselhaus soi
modify the electron @xmath1-factors in a si qd and enable the stark shift of the @xmath1-factors through gate voltage tuning@xcite .
different sign of the rashba ( @xmath8 ) and dresselhaus coefficients ( @xmath9 ) results in different g - factors among the two valley states@xcite .
the dresselhaus contribution is usually much stronger than rashba soi@xcite and dominates the @xmath1-factor renormalization @xcite .
[ fi1 ] in a si qd under strong vertical electric field , the electron is usually confined to only one interface . a monoatomic shift in the location of this interface results in a sign inversion of the dresselhaus coefficient ( @xmath10 ) but
the rashba coefficient ( @xmath11 ) remains unchanged@xcite . in practice ,
si / sige or si / sio@xmath0 interface certainly contains roughness , i.e monoatomic steps@xcite .
a non - ideal interface with monoatomic steps can be thought of as multiple smooth interface regions , where interfaces of neighboring regions are shifted by one atomic layer with respect to each other .
thus the neighboring regions will result in an opposite sign of @xmath10 .
an electron wavefunction spread over multiple such regions will witness multiple local @xmath10s and the effective @xmath10 will be an weighted average of these @xmath10s .
thus the presence of interface steps can change both the sign and magnitude of the effective dresselhaus contribution to the electron @xmath1-factors in a si qd@xcite . here
we show from atomistic tight - binding simulations of a si qd how the @xmath1-factors and their stark shifts can vary due to the presence of monoatomic steps at the interface . such variations in the @xmath1-factors and their gate voltage dependence is measured in gate defined quantum dots formed in a si / sio@xmath0 interface for two different samples ( a and b ) with similar architecture . in sample
a , we were able to form three neighboring qds and in sample b , we were able to form only one qd .
interestingly , there are significant differences in the measured @xmath12 times among one - electron and three - electron qubits realized in these dots .
we understand that the one - electron qubit corresponds to electron occupying the lower energy valley state @xmath3 , whereas the three - electron qubit corresponds to electron occupying the higher energy valley state @xmath4 . finally with the help of both atomistic and analytic calculations ,
we show that by changing the direction of the external magnetic field we can control the variability and stark shift of the g - factors and improve the dephasing times , @xmath12 .
[ fi2 ] fig .
1 shows how a monoatomic step at the interface of a si qd can affect the @xmath1-factors of the valley states and their electric field dependence , with an external magnetic field along the [ 110 ] crystal orientation , from atomistic tight - binding simulations .
an electron wavefunction near an interface step is shown in fig .
1a .
the distance between the dot center and the location of the edge of the interface step is denoted by @xmath13 .
the dot radius is around 10 nm .
so for @xmath14 - 10 nm the dot is completely on the left side of the step and has different @xmath1-factors compared to that for @xmath1510 nm , when the dot is completely on the right side of the step as seen in fig .
1b .
for -10 nm@xmath1610 nm , the @xmath1-factors are weighted average of those of the two sides based on the dot location .
to understand this atomistic calculation we look into an analytic model that relates @xmath17 in a si qd , with the rashba and dresselhaus soi@xcite .
for @xmath5 along the [ 110 ] crystal orientation @xmath18}=-2\frac{\left| e \right|\left\langle z \right\rangle } { \mu_\textup{b } \hbar } { { \alpha } _ { \pm } } + 2\frac{\left| e \right|\left\langle z \right\rangle } { \mu_\textup{b } \hbar } { { \beta } _ { \pm } } \label{equ4}\end{aligned}\ ] ] here , @xmath19 is the electron charge and @xmath20 is the spread of the electron wavefunction along the vertical direction z ( [ 001 ] ) .
now , in a si qd , @xmath21@xcite , @xmath18}\approx2\frac{\left| e \right|\left\langle z \right\rangle } { \mu_\textup{b } \hbar } { { \beta } _ { \pm } } \label{equ5}\end{aligned}\ ] ] as previously discussed , @xmath10 has a different sign between the two sides of the step .
when the location of the dot changes with respect to the step , the weighted average of the positive and negative @xmath10s change , which changes the g - factors .
figs .
1c and 1d show that the stark shift of the @xmath1-factors as a function of @xmath22 for both valley states are also affected by the presence of an interface step .
the change in the @xmath1-factors with electric field @xmath23 can change in both sign and magnitude based on the location of the step with respect to the dot center .
this behavior can also be explained from the change in the dresselhaus effect with an interface step and from equation [ equ5 ] .
for example in fig .
1c , for @xmath24 -10 nm , the dot is completely on the left side of the step , where the @xmath3 valley state has positive @xmath10 .
thus an increase in @xmath25 with increasing @xmath22 increases the @xmath26 as well , hence a positive @xmath27 . on the other hand ,
when the dot is completely on the right side of the step , at @xmath24 10 nm , @xmath25 is negative .
thus increasing @xmath22 increases @xmath28 but decreases @xmath26 and thus results in a negative @xmath27 .
for -10 nm@xmath1610 nm , @xmath27 changes gradually with @xmath13 .
we see a similar but opposite change for @xmath29 in fig . 1d .
[ fi3 ] in fig .
2 we show the schematic diagram of the devices measured experimentally and the observed variations in the @xmath1-factors and their stark shifts for two different samples . in sample
a , we observe variations in one - electron ( @xmath3 ) and three - electron ( @xmath4 ) @xmath1-factors among three neighboring qds ( @xmath30 , @xmath31 , @xmath32 ) , as shown in fig .
2b .
we could not achieve three - electron spin resonance for @xmath31 as it is strongly coupled to the other dots .
fig .
2c shows the stark shift of @xmath30 on sample a and one qd on sample b. we see here that the @xmath26 of @xmath30 has opposite dependence on @xmath22 compared to that of the qd in sample b. these observed variations in stark shifts and also in the @xmath1-factors qualitatively agree with the theoretically predicted variations shown in fig . 1 . thus we understand that these experimentally observed variations are due to different interface conditions associated with each of the qds .
we also observe variations in the measured @xmath12 times for both valley states of the three qds as shown in fig .
3 .
the @xmath12 times shown in fig .
3f were extracted by performing ramsey experiments on all of the qubits in sample a , as shown in figs .
3a - e .
the observed variations in @xmath12 can be explained from the measured ( fig .
2c ) and predicted ( fig .
1c , d ) variations in @xmath33 with interface steps .
the larger the @xmath33 , the more sensitive the qubit is to charge noise .
the dephasing time due to nuclear spin fluctuation is given in ref .
@xcite . in absence of nuclear spin
, we can relate @xmath12 times with charge noise in a similar way , @xmath34 here , @xmath35 is an effective fluctuation in the electric field seen by the dot due to charge noise and @xmath36 is the reduced planck constant .
when we compare the @xmath12 times between the two valley states of @xmath30 , we see @xmath37 @xmath38 and from ref .
@xcite we find , @xmath39 @xmath40 .
this comparison also shows how @xmath41 almost linearly depends on @xmath33 as predicted by equation [ equ1 ] .
thus variations in @xmath33 due to different interface conditions among different qubits will result in a variability in the dephasing time .
adversely , the performance of a large scale quantum computer might be limited by the performance of the qubit that decoheres the fastest .
the calculations of fig . 1 and the experimental observations of figs . 2 and 3 bring forward the device - to - device variability issues that would require individual knowledge of each unique qubit and can impose a challenge to the implementation of a large scale quantum computer .
any possible way of reducing the variability in @xmath17 , @xmath23 and @xmath12 is crucial to scale up the si qd spin qubits .
also an increase in @xmath12 , regardless of the interface condition , will be a large step towards scalability .
next we investigate ways to improve these issues . one obvious way to suppress these variabilities and gain more control over the g - factors , their tunability and the dephasing times is to minimize interface roughness , which is a well known fabrication challenge .
here we propose an alternate way to control these variabilities .
as predicted in ref .
@xcite , the @xmath1-factors in a si qd are anisotropic .
we can study the anisotropy from a simplified expression@xcite , @xmath42 here , @xmath43 is the angle of the external magnetic field with the [ 100 ] crystal orientation . from equation [ equ2 ]
we see that the contribution of the dresselhaus soi is anisotropic .
thus by changing the direction of @xmath5 we can tune the dresselhaus contribution . for @xmath44 , the dresselhaus contribution will be negligible .
for example , when @xmath5 is along [ 100 ] , @xmath45 and @xmath46}\approx-2\frac{\left| e \right|\left\langle z \right\rangle } { \mu_\textup{b } \hbar } { { \alpha } _ { \pm } } \label{equ3}\end{aligned}\ ] ] comparing equation [ equ5 ] and [ equ2 ] we see that , @xmath47}}{\delta g_{\pm}^{[110]}}\approx\frac{{\alpha } _
{ \pm } } { { \beta } _ { \pm } } \label{equ6}\end{aligned}\ ] ] as the effect of the monoatomic steps are more dramatic on @xmath9 , the change in @xmath17 and @xmath23 with interface steps should be smaller for [ 100 ] compared to that for [ 110 ] . moreover , since @xmath48 @xcite , @xmath23 itself will be much smaller for [ 100 ] .
[ fi4 ] in fig .
4 we show the changes in the @xmath1-factors with an interface step and electric field for @xmath5 along [ 100 ] . figs .
4a and 4b show @xmath26 and @xmath29 respectively as a function of @xmath22 for different values of @xmath13 .
it can be seen that the range of change with both @xmath22 and @xmath13 is much smaller compared to that in figs .
1c and 1d , when @xmath5 is along [ 110 ] . in fig .
4c we compare the change in @xmath1-factors between [ 100 ] and [ 110 ] , with respect to @xmath13 for @xmath4928.5 mvm@xmath50 .
though there are variations in @xmath26 and @xmath29 due to the change in the location of a dot from an interface step along [ 100 ] , as shown by the inset of fig .
4c , these variations are negligible compared to that along [ 110 ] . also @xmath23 is negligible , with a negligible variation with @xmath13 , along [ 100 ] compared to that along [ 110 ] as shown in fig .
4d .
such phenomenon is critical in a large scale quantum computer made of si qds .
if the external magnetic field is along the [ 100 ] crystal orientation , all of the qubits will have negligible variations in @xmath17 and @xmath23 among them even with their varying interface conditions .
very small @xmath33 along [ 100 ] would result in very large @xmath12 times . from fig .
4d , we calculated the average @xmath33 over different values of @xmath13 and find that , @xmath51\right)|\right)\approx30 $ ] @xmath52\right)|\right)$ ] and @xmath53\right)|\right)\approx50 $ ] @xmath54\right)|\right)$ ] .
this large decrease in @xmath33 can significantly increase @xmath12 times .
if a linear relationship between @xmath41 and @xmath33 holds as predicted by equation [ equ1 ] , the @xmath12 times can be larger than 1 ms , provided other noise sources do not contribute significantly .
now , a decrease in @xmath33 might mean a reduced tunability of the @xmath1-factors which is necessary for selective addressing of individual qubits .
but an increase in @xmath12 times will result in a narrower electron spin resonance ( esr ) linewidth , @xmath55@xcite , which would now require smaller difference in @xmath1-factors between qubits to individually address them .
as the @xmath1-factors and @xmath23 have lesser variations with interface roughness along [ 100 ] , the @xmath1-factor difference between qubits can be more controlled .
also a large enough @xmath12 time might allow for slower rabi oscillations and electric dipole spin resonance ( edsr ) might be achieved using just soi in a si qd , without the necessity of micro - magnets@xcite .
moreover , by controlling the direction of the external magnetic field a desired trade off between large @xmath33 and large @xmath12 times can be achieved . to conclude
, we experimentally observe variations in the electron @xmath1-factors and their sensitivity to electric field , which also leads to variability in the dephasing times among si / sio@xmath0 qds .
we identify that the source of the randomness is the presence of monoatomic steps at the interface . to gain control over these key qubit parameters one has to minimize interface roughness , which is challenging to achieve .
we have shown here that even in the presence of interface steps we can control and minimize these variations by taking advantage of the anisotropic soi in a si qd .
importantly , we can increase @xmath12 times if we align the external magnetic field along the [ 100 ] crystal orientation , rather than along [ 110 ] , where @xmath12 would be minimum . pointing the external b - field along [ 100 ]
will also help to reduce the soi induced dephasing in si qd devices with integrated micro - magnets , as soi also contributes to the @xmath1-factors in these devices even in the presence of micro - magnets@xcite .
100 vandersypen , l. m. k. _ et al . _ interfacing spin qubits in quantum dots and donors - hot , dense and coherent
. _ arxiv : _ 1612.05936 ( 2016 ) loss , d. & divincenzo , d. p. quantum computation with quantum dots .
_ phys .
rev . a _ * 57 * , 120126 ( 1998 ) . zwanenburg , f. a. _ et al . _
silicon quantum electronics . _ rev .
mod .
phys . _ * 85 * , 9611019 ( 2013 ) .
itoh , k. m. & watanabe , h. isotope engineering of silicon and diamond for quantum computing and sensing applications . _ mater .
res .
soc .
commun . _ * 4 * , 143157 ( 2014 ) maune , b. m. _ et al .
_ coherent singlet - triplet oscillations in a silicon - based double quantum dot .
_ nature _ * 481 * , 344347 ( 2011 ) .
wu , x. _ et al .
_ two - axis control of a singlet - triplet qubit with integrated micromagnet .
_ proc .
natl .
acad .
sci .
usa _ * 111 * , 11938 ( 2014 ) .
eng , k. _ et al .
_ isotopically enhanced triple quantum dot qubit .
_ science advances _ * 1 * , no .
4 , e1500214 ( 2015 ) .
kim , d. _ et al .
quantum control and process tomography of a semiconductor quantum dot hybrid qubit .
_ nature _ * 511 * , 70 - 74 ( 2014 ) .
veldhorst , m. _ et al . _ an addressable quantum dot qubit with fault tolerant control fidelity .
_ nature nanotechnology _ * 9 * , 981985 ( 2014 )
. kawakami , e. _ et al .
_ electrical control of a long - lived spin qubit in a si / sige quantum dot .
_ nature nanotechnology _ * 9 * , 666670 ( 2014 ) .
veldhorst , m. _ et al . _ a two - qubit logic gate in silicon .
_ nature _ * 526 * , 410414 ( 2015 ) .
zajac , d. m. , hazard , t. m. , mi , x. , nielsen , e. & petta , j .
r. scalable gate architecture for a one - dimensional array of semiconductor spin qubits .
_ phys .
rev .
appl . _ * 6 * , 054013 ( 2016 ) .
jones , c. _ et al . _ a logical qubit in a linear array of semiconductor quantum dots .
_ arxiv : _ 1608.06335 ( 2016 ) .
veldhorst , m. , eenink h.g.j . , yang , c.h .
& dzurak , a.s .
silicon cmos architecture for a spin - based quantum computer .
_ arxiv : _ 1609.09700 ( 2016 ) .
yang , c. _ et al .
_ spin - valley lifetimes in a silicon quantum dot with tunable valley splitting .
_ nature communications _ * 4 * , 2069 ( 2013 ) .
veldhorst , m. _ et al .
_ spin - orbit coupling and operation of multivalley spin qubits .
_ phys .
rev .
b _ * 92 * , 201401(r ) ( 2015 ) .
scarlino , p. _ et al .
_ second - harmonic coherent driving of a spin qubit in sige / si quantum dot .
_ phys .
rev .
lett . _ * 115 * , 106802 ( 2015 ) .
scarlino , p. _ et al .
_ , dressed photon - orbital states in a quantum dot : inter - valley spin resonance . _
arxiv : _ 1608.06538 ( 2016 ) .
ferdous , r. _ et al .
_ valley dependent anisotropic spin splitting in silicon quantum dots .
_ arxiv : _ 1702.06210 ( 2017 ) .
golub , l. e. , & ivchenko , e. l. spin splitting in symmetrical sige quantum wells .
_ phys .
rev .
b _ * 69 * , 115333 ( 2004 ) nestoklon , m. o. , golub , l. e. & ivchenko , e. l. spin and valley - orbit splittings in sige / si heterostructures . _ phys . rev . b _ * 73 * , 235334 ( 2006 ) .
nestoklon , m. o. , ivchenko , e. l. , jancu , j. -m . & voisin , p. electric field effect on electron spin splitting in sige / si quantum wells .
_ phys .
rev .
b _ * 77 , * 155328 ( 2008 ) .
zandvliet , h. j. w. & elswijk , h. b. morphology of monatomic step edges on vicinal si(001 ) .
_ phys .
rev .
b _ * 48 , * 14269 ( 1993 ) . friesen , m. , eriksson , m. a. & coppersmith , s. n. magnetic field dependence of valley splitting in realistic si / sige quantum wells , _ appl
. phys .
lett .
_ * 89 * , 202106 ( 2006 ) .
kharche , n. , prada , m. , boykin , t. b. & klimeck , g. valley splitting in strained silicon quantum wells modeled with 2 miscuts , step disorder , and alloy disorder .
_ appl .
phys .
lett . _ * 90 , * 092109 ( 2007 ) . merkulov , i. a. , efros , al . l. & rosen , m. electron spin relaxation by nuclei in semiconductor quantum dots .
_ phys .
rev .
b _ * 65 , * 205309 ( 2002 ) .
hanson , r. , kouwenhoven , l. p. , petta , j. r. , tarucha , s. , vandersypen & l. m. k. spins in few electron quantum dots . _ rev .
mod .
phys . _ * 79 * , 1217 ( 2007 ) . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.